Performance standards for the Food Stamp Employment and Training Program.

# Performance Standards for the Food Stamp Employment and Training Program

February 1992

OTA-ITE-526

NTIS order #PB92-157932

## Recommended Citation:

U.S. Congress, Office of Technology Assessment, 

## Performance Standards for the Food Stamp Employment and Training Program, OTA-ITE-526 (Washington, DC: U.S. Government Printing Office, February 1992).

For sale by the U.S. Government Printing Office Superintendent of Documents, Mall Stop: SSOP, Washington, DC 20402-9328

ISBN 0-16 -036121-4

# Foreword

The Food Stamp Employment and Training Program (FSET) was designed to reduce food stamp outlays by increasing the employment and earnings of able-bodied food stamp recipients. The States operate FSET with funding and guidance from the Food and Nutrition Service (FNS) of the U.S. Department of Agriculture.

To hold the States accountable for their implementation of FSET, Congress directed FNS to develop performance standards in the Hunger Prevention Act of 1988 (P.L. 100-435). The law directs FNS to develop standards based on outcomes, such as job placements, and to offer the States financial rewards and sanctions, based on their attainment of these standards.

As required by the Hunger Prevention Act, FNS proposed its performance standards for FSET on August 30, 1991. This report, 

## Performance Standards for the Food Stamp Employment and Training Program, was also mandated in the Hunger Prevention Act. As required by the act, this report describes OTA’s model performance standards and compares OTA’s model standards to those proposed by FNS.
However, measuring the effectiveness of FSET, or any employment and training program, requires that a distinction be made between outcomes and impacts. Although employment is the desired 

## outcome of such programs, many welfare recipients find employment on their own. Studies using an experimental design show that some employment and training programs have had a significant, positive 

## impact, helping more welfare recipients find jobs than would have without assistance. However, a study by Abt Associates found that FSET had no impact on the earnin gs of food stamp recipients.
Because performance standards are based on outcomes, they do not show whether a program has in fact increased employment and earnings beyond that which would have occurred without it. Thus, no performance standards, neither those proposed by FNS nor OTA’s model standards, can measure whether State FSET programs are having an impact. This report, then, goes beyond the original mandate and analyzes successful employment and training programs. Based on this analysis, the report identifies several alternative approaches to increasing the impact of FSET.

/ff~ 

# 64A# ‘ >

JOHN H. GIBBONS
Director

. . . Ill

# Reviewers—Performance Standards for the Food Stamp Employment and Training Program

Steven Aaronson, Chief, Adult and Youth Standards Unit U.S. Department of Labor Employment and Training Administration Washington, DC

Janet Baldwin Assistant Director for Policy Research GED Testing Service American Council on Education Washington, DC

Hal Beader Graduate School of Education Rutgers University Princeton, NJ
Stephen H. Bell, Senior Economist Abt Associates Bethesda, MD

Stephen Cameron National Opinion Research Center The University of Chicago Chicago, IL
Lela J. Dyson Employment Specialist Policy and Program Design Department Texas Department of Human Services Austin, TX

Steve Carlson Director, Office of Analysis and Evaluation Food and Nutrition Service U.S. Department of Agriculture Alexandria, VA

Daniel Friedlander Manpower Demonstration Research Corp. New York, NY

Barbara S. Goldman Vice President for Research Manpower Demonstration Research Corp. New York, NY
Gary Gortenberg Bill Hastings Policy Consultant Family Independence Division State of Indiana Department of Public Welfare Indianapolis, IN
Ellen Hennigan Supervisor, Work Program Section Food and Nutrition Service U.S. Department of Agriculture Alexandria, VA

Sar Levitan Center for Social Policy Studies Washington, DC
Larry McDowell Administrator, Social Services Adult Services Unit Jefferson County Department of Social Services Golden, CO

Mary Ellen Natale Food and Research Action Center Washington, DC
Russ Overby Attorney Legal Services of Middle Tennessee Nashville, TN
Ron Pugsley Chief, Program Services Branch Division of Adult Education and Literacy U.S. Department of Education Washington, DC

Michael J. Puma Vice President Abt Associates Bethesda, MD

Program Manager for Employment and Training Fairfax County Fairfax, VA

NOTE: OTA 

## appreciates and is grateful for the valuable assistance and thoughtful critiques provided by the reviewers. The reviewers do not, however, necessarily approve, disapprove, or endorse this report. OTA assumes full responsibility for the report and the accuracy of its contents. iv

# OTA Project Staff—Performance Standards for the Food Stamp Employment and Training Program

Lionel S. Johns, 

## Assistant Director, OTA Energy, Materials, and International Security Division

Audrey Buyrn, 

## Manager, Industry, Technology, and Employment Program

Margaret Hilton, 

## Project Director

Gretchen Kolsrud, 

## Senior Associate

Contributors

Linda Roberts
Toby Rogers



## Administrative Staff

Carol A. Guntow, Office 

## Administrator

Diane D. White, 

## Administrative Secretary

Publishing Staff

Mary Lou Higgs, 

## Publishing Manager

Denise Felix

Christine Onrubia
Bonnie Sparks

Susan Zimmerman

# Contents

Page



## Section 1. Introduction and Summary ... m.. ... ... .m. ... .*. **** ... .**. .m. ..** *m*. *****.0 INTRODUCTION

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . SUMMARY

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Section 2. History and Context of FSET

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Section 3. Issues in Setting Performance Standards

● 

## *.m. ... ... ... ... .**. .. **0********..

Section 4. Comparison Between FNS’ Proposed Standards

and OTA’s Model Standards . . . . . . . . ., . . . . . . . . . . . . . . . . . . . . . . . . . . . . .....0...
CRITERIA FOR DEVELOPING STANDARDS

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . FNS AND OTA STANDARDS

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Participation Rate

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Job Placements

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. Earnings

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Reductions in Food Stamp Dependency

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Educational Improvements

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Calculation of Overall Performance

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Section 5. Policy Options .o. .*o. ... .*. ... ... ..o. ... ... e.. ..*. ... *.. *.. *as*.**..**.*.**** OPTION 1: MAKE FSET PARTICIPATION VOLUNTARY

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . OPTION 2: MERGE FSET INTO JTPA

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1 1

5

7



## 11
11 12 12 16 18 20
21 28

31 31 32



## Boxes

Box Page 

## A. What Is the GED? . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 B. What Kind of Education Is Most Likely To Help FSET Participants? . . . . . . . . . . . . . . . . . . . . 26



## Figures Figure Page 

## 1. A Small, Niche Program

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. Per Capita Expenditures . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. Food Stamp Recipients and FSET . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. Subgroup Impacts

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .



## Tables 5
6 8 16

Table Page 

## 1. Performance Measures Selected-Comparison Between FNS’ Proposed Rule and OTA’s Model Standards

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. Comparison of Initial National Standards

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. Costs and Earnings , Impacts of Mandatory Employment and Training Programs for AFDC Recipients

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. Job Placement Rate-Comparison Between FNS’ Proposed Rule and OTA’s Model Standards . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5. Average Wage and Earnings-Comparison Between FNS’ Proposed Rule and OTA’s Model Standards

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6. Assessment of Overall Performance: Comparison of FNS’ Proposed Rule and OTA’s Model Standards

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 12

14

17

20

29

# Section 1 

# Introduction and Summary

INTRODUCTION

This report on proposed performance standards for the Food Stamp Employment and Training Program (FSET) responds to a mandate in the Hunger Prevention Act of 1988 (Public Law 100- 435). That law directs the Secretary of Agriculture to develop new, outcome-based performance standards for assessing State implementation of FSET, inconsultation with the States, the Office of Technology Assessment (OTA) and other Federal agencies. The law also directs OTA to report to Congress within 180 days of publication of the proposed performance standards, l 

## outlining model performance standards for FSET and comparing those models with the standards proposed by the Food and Nutrition Service (FNS) of the U.S. Department of Agriculture (USDA).

FNS published the proposed performance standards on August 30, 1991. Subsequently in the Food, Agriculture, Conservation and Trade Act of 1991 (Public Law 102-237, signed December 13, 1991), Congress allowed FNS to delay implementation of final performance standards until 1 year after the U.S. Department of Health and Human Services (DHHS) publishes final performance standards for its Job Opportunities and Basic Skills (JOBS)

program. 2 

## Since DHHS is unlikely to publish final performance standards until 1994 or 1995, 3 

## FNS and Congress will have ample time to consider OTA’s suggestions and policy options.

In preparing this report, OTA drew on a base of knowledge developed over the past eight years, starting with its study of retraining and reemploying displaced workers, published in 1986. 4 

## More recently, OTA published a major assessment ofworker training and its impacts on U.S. competitiveness. Analysis of employment and training was alsocritical to OTA assessments of international competition in manufacturing and in services. Finally, OTAstaff involved in education studies provided valuable background information. As part of the study, OTA participated in a series of meetings of an advisory panel convened by USDA to assist in developing the FSET performance standards. The OTA-USDA relationship has been cordial and cooperative.
This report has five sections:

1. 

## 2. 3.

4.

5. This Introduction and Summary; History and Context of FSET;Issues in Setting FSET Performance Standards;Comparison Between FNS’ Proposed Standards and OTA’s Model Standards; and Policy Options.



# SUMMARY

FSET occupies a niche between two much larger Federal employment and training programs-the Job Training Partnership Act (JTPA) and the Job Opportunities and Basic Skills Program (JOBS). The goal of FSET is to reduce food stamp outlays byincreasing the employment and earnings of ablebodied food stamp recipients.

A comprehensive evaluation of first-year implementation of FSET conducted by Abt Associates, Inc. found that the program was not meeting this goal. OTA concurs with Abt’s conclusion that drastic change is needed if this goal is to be met. However, OTA finds that performance standards, by themselves, are inadequate to accomplish such a change. FNS has proposed two alternative models to implement the performance standards. 5 

## Model A is

156 Federal Register 43152 (Aug. 30, 191)

2JoBs is ~ ~ploPent ~d ~- progr~ for welf~e 

## r~ipients, &SCUSS~ tier in section 2 Of hk report. s~eF@y su~rt Act of 1988 fiblic Qw 1~85) directs D~S to develop pelfO rmance standards for JOBS and submit its recommendations to Congress by October 30, 1993.

4u.s. con~ess, Offlce of ~c~olou ~=ssmen~ worker Training: competing in the NeWI ~nt~national Economy, ()~-~57 (w*@OQ

DC: U.S. G OV ernment Printing OffIce, September 1990); Making Things Better.’ Competing in Manufacturing, O’IA-ITE-443 (Washington, DC: U.S. Government Printing Office, February 1990); 

### International Competition in Smices: Bank% Buildings so~are~ Know-How...P OTA-~E-32$ (Springfield, VA: NTIS, July 1987); Technology and Structural Unemployment: Reemploying Displaced Adults, OTA-ITE-250 

### (Washi.ngto& DC: U. S. Government Printing OffIce, February 1986)

S56 Federal Register (Aug. 30, 1991), p. 43152.



## –l–

### 2 ● 

### Performance Standards for 

## the Food Stamp Employment and Training Program



## similar to the approach used in JTPA and JOBS. It requires the States to gather data on participant outcomes, such as the percent of participants that find jobs. 6 

## Model B requires States to gather information on outcomes for both FSET participants and control groups of nonparticipants. Outcomes for the two groups would be compared to determinewhether there was a statistically significant difference. Both models would allow FNS to distinguish between the States and award incentive funds to those States whose programs were most successful in meeting national standards established by FNS.

OTA found that, although model B would providea much more accurate measure of the true effectiveness of each State’s FSET program, it would beimpractical to implement. OTA’s rationale is discussed further in section 3 below. OTA found that the performance standards proposed in model A would encourage the States to make improvements, but would not, by themselves, create the degree of change needed to make FSET effective in meeting its stated goals.

OTA agrees with two of the four performance measures proposed by FNS--job placements and educational improvements. OTA also supports FNS’ proposal that educational improvements remain an optional component of State FSET programs, since there is no firm evidence that remedial education, by itself, reduces welfare dependency and increasesearnings. However, OTA would define job placements and educational improvements somewhat differently than FNS and would use a differentweighting scheme to encourage placements in lasting jobs and service to those who are harder to employ.
To comply with the Hunger Prevention Act’s direction that the performance standards “take into account” both wages and job retention, OTA suggests using average quarterly earnings among those employed, rather than average hourly wages among those employed, as a third performance measure. OTA disagrees with FNS’ fourth measure-- food stamp case closures-because this measure could force those who are hard to employ off food stamps and out of FSET. Finally, both OTA and FNS conclude that establishing performance standards should bean ongoing process, and that the standards should be revised based on State experience.
FNS proposes the following initial performance standards for fiscal years 1992-94:

1. 2. 3. 4.

5. a participation rate of 10 percent; a job placement rate of 25 percent; an average wage of $4.45 per hour; a food stamp case closure rate of 20 percent; and educational improvements among 25 percent of FSET participants who enroll in educational

programs.OTA generally supports the participation standard, the job placement rate standard and the educational improvement standard. However, OTA would not use food stamp case closures as a performance measure, and suggests a quarterly earnings standard that corresponds to a slightly higher average wage rate of $5 per hour. OTA’s rationale for its support of FNS and for its areas of disagreement is discussed in section 4.

OTA concurs with FNS’ proposal that an adjustment model be used to vary the initial national standards to avoid penalizing the States for factorsbeyond their control, such as a very high unemployment rate or low average wages. OTA finds the model proposed by FNS reasonable.

OTA’s analysis of FSET suggests several policies that Congress may wish to consider. First, as notedabove, FNS’ proposed model B performance standards appear impractical to implement, but the more feasible model A standards are unlikely to measure the true impact of each State’s FSET program. Because of this limitation, Congress may want to reconsider its policy of linking State attainment of the performance standards to financial rewards and sanctions. More significant, however, is OTA’s conclusion that no outcome-based performance standards, including OTA’s own model standards, will, by themselves, bring about the degree of change needed to enable the program to meet its goals. Such change might be helped by two other policy options which would restructure FSET— making the program voluntary and/or merging it with JTPA.

%s model also allows FNS to adjust the standards to account for factors beyond the State’s control. For example, a State with a high unemployment rate might be held to a lower job placement standard.

### Section 1--Introduction and Summary ● 

## 3

OTA found that mandating participation in FSET has contributed to the program’s lack of impact. Congress may wish to consider making FSET voluntary. If this option is chosen, performance standards will need strong positive incentives to encourage the States to recruit less-employable volunteers as well as those who are job-ready. Another policy option that could reduce administrative overhead and increase direct provision of employment and training services would be to merge FSET with JTPA. Many States and localities are already using JTPA as their primary service provider, either through formal contracts with their JTPA agency or through informal referrals.

# Section 2 

# History and Context of FSET

Figure 1—A Small, Niche Program



## I 1046 I

1000

800

w 400

200

0
JTPA-IIA FSET JOBS

program 

### Total FY 1990 Federal expenditures for three employment and training programs.
SOURCES: U.S. General Accounting Office, Training Programs: /formation on Fisml

Years 1989 and 1990 Appropriations, pp. 7,9, 11, and Congressional Budget Office, OTA, 1991.



## FSET has evolved out of a 20-year effort to reduce food stamp dependency by encouraging food stamp recipients to work. Congress amended the Food Stamp Act in 1970 (Public Law 91-671) to require all able-bodied adult recipients to register for work with their local Employment Service office, and to accept employment if offered. During the 1970s andearly 1980s, USDA helped some States and localities operate demonstration programs involving work experience and job search training for those food stamp recipients who were required to register for work (’‘work registrants’ ‘). In the Food Security Act of 1985 (Public Law 99-198) Congress required all States to create employment and training programs for work registrants, and provided funding for thenew Food Stamp Employment and Training programs. These funds include both a basic grant andadditional Federal matching funds (on a dollar-fordollar basis) for States that invest their own money in supportive services such as transportation and child care for FSET participants.

Today, FSET occupies a niche between two, much larger, federally funded employment andtraining programs (see figure 1). The Job Opportunities and Basic Skills training program, or JOBS, 7 

## istargeted to mothers and unemployed fathers receiving Aid to Families with Dependent Children (AFDC), while Title IIA of the Job Training Partnership Act of 1982, or JTPA, 8 

## is available on a voluntary basis to a broad range of economically disadvantaged adults and youth. Total fiscal year (FY) 1990 Federal outlays for FSET (including both basic grants to the States and Federal outlays to match State funds) were $148 rnillion, 9 

## compared to $1.04 billion for JTPA Title II-A 1O 

## and $264 million for JOBS. ll 

## Both JOBS and FSET emerged from welfare reform, requiring welfare recipients to either work (in unpaid ‘‘workfare” public service jobs) or look for work as a condition of receiving benefits. Although the concept that welfare recipients owe a reciprocal obligation in exchange for their benefits has not changed, most State and local JOBS andFSET programs now require participation in employment and training programs, rather than in workfare.

Because FSET and JOBS are ‘‘mandatory’ programs, the States are required to enroll a large fraction of those eligible and to sanction those who do not enroll by reducing their welfare benefits. To meet the required participation rate, the States spread a total Federal budget of $148 million across 1.35 million FSET participants in FY 1990, leading to an average of only $110 per participant. 12 

## Federal expenditures for JOBS averaged $692 per person

~amily fhlpport Act of 1988 (FSA), Public htW 100485.

8Public Law 97-300.

%J.S. OffIce of Management and Budget “Budget of the United States Government FY 1992,” Washington DC, February 1991, Appendix.

I?Ibid.

1lu.s. ‘1’remury estimate, 

### reportti by Janice Pesl@ Congressionrd Budget Office, personal Communication% Sept. 19, IW1.

12~s

~~ ~it~ by J~e Isaacs, cogessio~ Budget offIce, personal communicatio~ Sept. 23, 1$)$)1. men state matching funds are included, total spending for FY 1990 was $221 rnilliom or $164 per participant.



### –5–

### 6 ● 

### Performance Standardsfor the Food Stamp Employment and Training Program



## that year. 13 JTPA 

## Title II-A, which is voluntary, enrolled 632,000 persons 14 

## with a budget of $1.04 billion, resulting in an average expenditure of $1,646 per participant (see figure 2).
There is little overlap between the populations served by JOBS and FSET—most mothers of young children are not required to, and in fact, do not, participate in FSET Among FSET participants in FY 1988, 74 percent lived in one- or two-person households without children. 15 

## However, there is agreater overlap between the FSET and JTPA populations. Food stamp recipients are automatically considered ‘economically disadvantaged’ and hence eligible for free employment and training services under JTPA. l6

JTPA’s existence 

## does not obviate the need for employment and trainin g services for food stamp recipients. Although JTPA has much more funding than FSET, it serves only a small fraction of those eligible for its services— 2.3 percent in 1986. 17 

## During the first half of program year 1990, 38 percent of the 632,000 JTPA Title IIA participants received food stamps. Of these, about half were able-bodied work registrants targeted by FSET. 18


## Assumin g these trends held through the year, JTPA 1

1

1

1 

### Figure 2—Per Capita Expenditures

800 1646 I

I

JTPA-II FSET JOBS

program Average Federal expenditure per participant for three employment and training programs, FY 1990. SOURCE: OTA, 1992.



## served about 120,000 food stamp work registrants. Creation of FSET with a congressional mandate to serve up to 50 percent of those eligible has providedemployment and training services to many additional food stamp recipients—about 1.35 million in FY 1990. 19 

## However, as discussed below, FSET has had little impact on the much larger group it serves. 20

13~ 1990 ~a~ ~ ~wition ~w for JOBS. me Smtes were not rqu~ed to begin operating JOBS progr~s IUIM

~ 1991. when fll~y OptXatiO@ JOBS is projected to require an average of $1,500 to $1,700 per participant-U.S. Congressional Budget OffIce, 

### “Work and Welfare: The FWY Support Aet of 19g8,” Washington DC, CBO Staff Working Paper, January 1989, p. 14.
IAEst~ted Progm yea 1990 emo~en~ based on e~~ents in be first

~ of tie ye~—u.s. Dep@ment of Labor, Employment ~d T-g Administration% Office of Strategic Planning and Policy Development, “Job Training Quarterly Survey: JTPA Title IIA and III Enrolhnents and
Te ruminations During the First Half of Program Year 1990 (July-December 1990),” 

### Washingto@ DC, July 1991, 

### p. 5.

15Mic~el J. ha, et ~., Evalmtion of th

e Fo~dStamP Employment and Training Program Final Report: Vozumez @ethesd~ MD: Abt Associates, Inc., June 1990), pp. 3-33-3-34.

16swtion 4(8) of JT’PA ‘f’ifle ~a defines “monomi~y disadv~~ged’ eligible individ~s to include individuals in fties receiving food ShlllpS.

17~s is the most recent&~ ~hich es~te p~icipation as a ~ction of ~ose leg~y eligible for J’TPA. It is from Steven H. Sandell ~d KRupp, Who Zs Served in JTPA Programs: Patterns of Participation and Intergroup Equity (Washi.ngtom DC: National Commission for Employment Policy, 1988), p. 50.

18u.so Dep~ent of ~~r, Division of p~o~~ce M~gement ~d Ev~wtio~ office of Strategic Pl- g and Policy Development “JTPA
Title IIA and III Enrollments and W ruminations During the First Half of Program Year 1990 

### (July-December 1990),’ 

### unpublished doc~ent Febmw 1991, p. 29.

19’IMS toti includes 1.2 million work registrants and 150,000 vOIUnt&rs.

20~s~ proposed new p~cipation Sadad of 10 percent could reduce total national tHMOhIdS tO M low as 2407W (one-f~ tie ‘~br ‘Wd under the current 50 percent participation standard).

# Section 3 

# Issues in Setting Performance Standards

Congress clearly desired the States to be held accountable for their success (or failure) in operating FSET. In the Food Security Act of 1985, Congress not only authorized State grants beg inning at $40 million in FY 1986 and growing to $75 million by FY 1989, but also directed USDA to establish Stateperformance standards designating minimum percentages of mandatory participants to be enrolled in FSET and allowing USDA to withhold funds fromany State that failed to enroll its required minimum. 21 

## That law specifies that the minimum percentage required of the States be no greater than 50 percent. In addition, the act required USDA to monitor the effectiveness of State implementation of FSET in terms of increased employment and job retention of participants and to report back to Congress on the program’s effectiveness. 22



## To implement the first part of this law, USDA required the States to either enroll or initiate sanctions against 35 percent of ‘non-exempt’ work registrants in FY 1989 and 50 percent in FY 1990 and FY 1991.

As shown in figure 3, even before USDA set these minimum percentages, in FY 1988, the States either enrolled or sanctioned 50 percent of ‘non-exempt’

or ‘‘mandatory’ work registrants. However, the total number enrolled or sanctioned-1.2 million— equals only about 4 percent of the 28 million persons who used food stamps that year. This is because the law exempts the vast majority of food stamp recipients from the work registration requirement, either because they are caring for young children, are under 17 or over 59, or already have work. 23



## Only 13 percent (3.7 million) of all food stamp recipients were classed as ‘‘work registrants’ in FY 1988, and USDA allowed the States to categorically exempt about one-third (1.2 million) of these from FSET participation. Categorical exemptions were allowed on the basis of geography (living in a remote area lacking an FSET program) and for other reasons, such as being in a household with three or more children. The remaining 68 percent of work registrants (2.5 million persons) were considered ‘‘mandatory’ participants, and the states enrolled or sanctioned nearly half of these.

Since establishing a minimum participation rate, FNS has encouraged the states to limit the total number of both categorical and personal exemptions to no more than 30 percent of all work registrants in each State. 24 

## The remaining 70 percent form the base of ‘‘non-exempt mandatories." The States were required to enroll 50 percent of this group in FY 1990.

To implement these participation-based performance standards FNS allocated $15 million, or 20 percent of the total $75 million in basic State grants for fiscal years 1990 and 1991 on the basis of each State’s attainment of the standards in previous fiscal

years.
The Hunger Prevention Act of 1988 called for new, outcome-based performance standards, and directed USDA to develop a proposal for modifyng State grant levels depending on how effective the States are at meeting these standards. Most recently, in the Food, Agriculture, Conservation and Trade Act of 1990 (Public Law 101-624), Congress directed USDA to reserve $15 million of the $75 million total authorized for State grants. During fiscal years 1992 through 1995, this $15 million is to be allocated among the States based on their performance, as measured by the new performance standards called for in the Hunger Prevention Act. The remaining 

## $60 million is to be allocated on the basis of the number of work registrants in each State. 25



## Although the question of how effective the States are relative to each other is an important one, it may

21 Actual tiding for tie b~ic sate grants was $50 million in FY 1987,$60 million in FY 1988, and $75 million in FY 1989 and 1990.

~ addition to the basic grants, USDAmatches State expenditures for transportation and child care, bringing total Federal expenditures forFY 1990 to $148 million.

~’rhe Food SeCW-@ Act of 1985 (Public Law 99-198), Title XV, 

### Subtitle A sec. 1517.
23- et aL, op cit, footnote 15, p. 3-21.

~Rur~ Stites me wowed to exempt a much higher percent of work registrants, but most states me held to a maximum of 30 to 35 percent. SOURCE: Ellen Heningeq Supervisor, and Nancy Theodore, Work Program Sectioq FNS, personal communication Dec. 5, 1991.

%3x. 1753.

– 7

–

### 8 ● 

### Performance 

## Standards for the Food Stamp Employment and Training Program

Figure 3-Food Stamp Recipients and FSET

Exclude 24.5 million over 59, under 18, or caregivers Y 

# & 3.7 million work registrants

1.2 million categorical exemptions (e.g., rural)

& 2.5 million mandatory FSET

1.3 million no show or further exemption



# o 1.2 million actual FSET (enrolled or sanctioned)

About 1 in 25 food stamp recipients participate in food stamp education and training (FSET). Most food stamp recipients are exempted from FSET because of age, their responsibilities as caregivers, or other circumstances such as living in a rural area with limited access to an FSET program. (It should be noted that since “sanctioning” counts as participation, substantive participation is less.)
SOURCE: OTA, using data for FY 19SS provided by FNS.



## not be as important as the question of how effective FSET is as a national program. To answer that question, USDA contracted with Abt Associates to conduct a 4-year, $3.5 -million study of FSET implementation in FY 1988, its first full year of operation. The study found that FSET had “nodiscernible effect on participants’ aggregate earnings, probability 

## of finding work, amount of time worked, or average wages. ” 26



## The authors of the Abt evaluation attributed the low impact partly to low participation levels. They found that about half of the “mandatory” FSET participants were never actively involved in the program, either because they failed to appear for their initial assessment interview (34 percent of those assigned to FSET), or because they met with
their caseworker and were determin ed exempt (15 percent of those assigned), or because they neverappeared for services after completing the assessment interview (3.5 percent of those assigned to the program). 27 

## The Abt researchers also speculated that, among those who did participate, employment

M- et al., op. cit., footnote 15, p. xi

Wbid., pp. 6-8.

### Section 3--Issues in Setting Performance Standards ● 

### 9



## and training services were unnecessary for one group and were spread too thinly across a second: the first, which was more job-ready, would have found jobs even without FSET; the second group, which had more serious barriers to employment, needed more intensive assistance to find jobs.

OTA concurs with Abt’s conclusion that drastic change is needed if FSET is to meet its goals. These include finding jobs for food stamp recipients,reducing the food stamp rolls, and ‘helping individuals to achieve self-sufficiency .” 28 

## Reaching these goals, and measuring progress toward them, may be difficult. The Abt study is conclusive because it usedan experimental design which allowed careful measurement of the true impact of FSET. FNS has asked for public comment on requiring the States to use either an experimental design (model B) or a nonexperimental design (model A) as the basis for performance standards.

There are many ways to study employment and training programs. For example, implementation studies are useful to find out whether or not a program is operating as planned. However, to evaluate the 

## effectiveness of employment and training programs, a distinction must be made between outcomes and impacts. Although job placements are the desired outcome of such programs, they may not represent a real impact. Many welfare recipients find jobs, whether or not they enroll in employment programs such as FSET. External factors (e.g., the status of the local economy) and internal factors (e.g., self-esteem) may affect an individual’s ability to find a job as much or more than participation in an employment and training program. In a recent study of performance standards for secondary school vocational education, OTA concluded that these confounding factors made ‘‘labor market indicators alone an insufficient basis for performance standards. ‘ ’ 29



## The best way to isolate the effect of employment and training programs from these other factors is to randomly assign like groups of individuals to the program and to a control group that does notparticipate and then compare the employment experience of the two groups. 30 

## The results can be surprising. For example, over 50 percent of those participating in FSET in FY 1988 had experienced some employment 1 year after entering the program, but so did a control group who did not enter FSET. 31



## FNS recognizes that an experimental design would be the ideal way to assess the true impact of each State’s FSET program. The model B alternativewould require the States to randomly assign mandatory FSET participants to treatment and control
groups. 32 

## The impact of FSET would be assessed through followup interviews with participants and controls, conducted at 6 months after random assignment.

However, FSET is a small program with a limitedbudget. Using random assignment is timeconsuming and expensive, primarily because program operators must be educated about the process to feel comfortable with turning away individuals who want to participate in order to form a control group. And some program operators are opposed to refusing services in order to create control groups.

When Abt Associates evaluated FSET, they encountered this problem. 33 

## Abt recruited State andlocal food stamp agency (FSA) directors to participate in the study through national and regional meetings. Both at these initial recruiting meetings and subsequently, some FSAs refused to participate, often because of their concerns about denying

zg~e Hunger prevention Act of 1988 (Public Law 100-435), Title IV, WC. 404, subsec. L (ii), I.

2~eS. conge~~, offIce of ~c~olog A~~essmen4 pe~omance Stan&-ds for seco&V school voea~”o~[ Education-1. ?ackgrouti Paper (Washingto@ DC: OTA, 1989), p. 5.
~uring the late 1970s and early 1980s, the Comprehensive Employment and Training Act (CETA) was evaluated by creating an artiflcird “conlrol group.” Social Security earnings records were merged with demographic and labor market data drawn from the sample of households surveyed each month through the Current Population Survey to create a control group whose characteristics matched those of CBTA participants. Although this approach is simpler than random assignmen~ the results are less reliable, because of the difilculty of matching the characteristics of participants and controls and assuring that controls do not receive program services. Most researchers now agree that random assignment is the best way to measure the impact of employment and trairdng programs. Sources: Burt S. Barnow, “The Impact of CETA Programs on Earnings: A Review of the Literature,” The JournaZofHumanResources, vol.

~ (2), p. 189; Gary Burtless and Larry Orr, “Are Classical Experiments Needed for Manpower Policy?” The Journal of Human Resources, vol. 21, No. 4, fall 1986, pp. 606-639; Robert LaLonde and Rebecca Maynard, “How Precise Are Evaluations of Employment and Training Programs,’ Evaluation Review, vol. 11, No. 4, August 1987, pp. 428-451.

31- et al., op. cit., footnote 15, p. A.

3256 Federal Register 43164 (Aug. 30, 1991).

33- et al., op. cit., footnote 15, pp. 4-9-4-15.

### 10 ● 

### Performance 

## Standards for the Food Stamp Employment and Training Program



## services to individuals assigned to the control group. Those FSAs that agreed to participate required extensive technical assistance: Abt staff conducted two rounds of site visits to each, wrote technical assistance manuals, hired on-site data collectors, and conducted staff training in order to assure that the random assignment and initial data collection on controls and participants went smoothly. Four years and $3.5 million were required to complete the study.
Similar problems could be expected if the States were required to use model B as the basis for their performance standards. The extra staff time and expense would be a burden on Federal and State FSET administrators, who already lack the fundingneeded to provide effective employment and training services.
FNS estimates the total paperwork burden on the States at 272,567 hours for model B, as opposed to only 187,859 hours under model A. However, this estimate specifically excludes the time needed for computer progr amming and operation and “the development and execution of the sampling and random assignment methodologies. ’ ’ 34 

## FNS anticipates that each State would have to assign a full-time person to oversee the random assignment process, and that, “ideally’ each locality would also assignone person to coordinate random assignment decisions and oversee data collection. 35 

## Assumin g that only the State-level staffing was required, this would add approximately 115,000 hours to the paperwork burden in model B, for a total of 387,567 hours. 36



## In addition to its high costs, use of control groups is impractical as long as FSET is, at least in theory, a mandatory program. For example, most work registrants in Los Angeles County are males who receive both State-funded General Assistance and food stamps. In return for these benefits, the State requires them to work in unpaid public service jobs. All are expected to perform this service, and sanctions are quickly applied to those who refuse. For this food stamp agency, assigning every other work registrant to a control group which would not be required to perform workfare would be out of the question.

OTA concludes that use of an experimental design to implement national performance standards is not feasible. 37 

## However, without random assignment, any performance standards will be imperfect indicators of the impact of the State programs.Because performance standards measure only outcomes and not the impacts of State programs, they provide a poor basis for financial rewards and sanctions. Based on its conclusion that model B is impractical, OTA’s discussion of FNS’ proposed standards in section 4 focuses on model A.

In addition to their limitations as measures ofprogram effectiveness, nonexperimental performance standards may have unintended consequences. For example, JTPA’s former performance standards, which emphasized m aximizing job placements while minimizing costs, encouraged some local programoperators to focus on the most job-ready participants, while providing minimal or no service towelfare recipients and others who were less employable. 38



## Recognizing these problems with performance standards, OTA has identified other policy options,which, if implemented, might enhance the effectiveness of FSET. These are discussed in section 5. Finally, OTA suggests that FNS conduct another comprehensive evaluation, perhaps in FY 1995, todetermine whether the proposed policies and performance standards, if implemented, are having the desired effect.

~56 Federal Register 43153 (Aug. 30, 1991).

3556 Federal Register 43164 (Aug. 30, 1991).

36Abt wed ~n.site &

~ collectors ~ site coordinato~ at 24 of he 53 local o&Iws involved in their Mtiond FSET evaluation study Abt Associates, op. cit., footnote 15, p. 4-15. When Manpower Demonstration Research Corp. (MDRC) used random assignment to evaluate the impact of State welfare-to-workprograms, it found that it needed at least one or two full-time staff people to collect and analyze the data fmmeachlocal office. However,
the Abt and MDRC were extensive, one-time studies. It is possible that fewer staff would be required if random assignment was used by FSET on an ongoing basis. Barbara Goldmaq Director of Researcb MDRC, personal communication April 1991.
sTOTArea~h~ a sw conclusion fi its @ysis of pe~omace stan~ds for secon~ school vocational education. OTA, 1989, Op. Cit., fOOtllOte 28, p. 5.
38~~e~e p. Dickemon, et & JTPA pe~o~n~e Standards: Effects on Clients, Semices and Costs (washingto~ DC: National COmlrl&SiOn fOr Employment Policy, 1988), p. 194.

# Section 4



# Comparison Between FNS' Proposed Standards

and OTA's Model Standards

OTA and FNS are in substantial agreement on many aspects of setting performance standards. For example, like FNS, OTA considered a set of overarching goals, or criteria, for establishing its standards. OTA agrees with FNS that, as long as FSET is mandatory, there must be a minimum participation standard. OTA further supports FNS’ proposal that the minimum participation rate be greatly reduced—horn 50 to 10 percent-in order to allow more intensive application of employment and training services to a smaller number of people. OTA also agrees with FNS’ selection of twooutcome-based performance measures--entered employment rate and educational improvements. However, OTA suggests using quarterly earnin gs of participants, rather than hourly wages, as one outcome-based measure. And, OTA disagrees with one measure proposed by FNS--the percentage of households that no longer receive food stamps. OTA supports FNS’ approach of setting an initial national standard as a benchmark, with future revisions to the standards as more data become available. OTA also agrees with two of the initial benchmarks proposed by FNS (see tables 1 and 2). However, there are a few areas of disagreement in addition to those noted above. These include the priority of service, some of the weights, and some of the initial benchmarks. Finally, as discussed in section 5, Congress and FNS may want to make some more fundamental changes to FSET.



# CRITERIA FOR DEVELOPING STANDARDS

In developing their proposed standards, both FNS and OTA examined legislation to assess whatCongress wanted from the new performance standards. Based on its analysis of the Hunger Prevention Act of 1988, FNS identified three goals, or “design

Table l—Performance Measures Selected-Comparison Between FNS’ Proposed Rule and OTA’s Model Standards

Measures considered FNS OTA

1.

2.

3.

4.

5. Participation rate Yes
Entered employment Yes
Average wage Yes

Reductions in food stamp dependency Yes

Educational

Job

Hard to employ (HTE) improvements Yes (optional)

Definition of terms Terminee (FNS) Those who are assigned to Participant (OTA) FSET, and either complete the program or drop out, excluding those who do not appear for their first activity

Must involve 

### 20 or more hours per week of unsubsidized employment and be expected to last at least 30 days

Participant who lacks a high school diploma/GED and has not worked in the prior year Yes--FSET is “mandatory”

Yes--legislative goal

Use earnings instead

No-may have unintended
consequences

Yes (optional--difficult to define

Same

Same

Same

SOURCE: OTA, 1992.

316-637 0 - 92 - 2 QL: 3 

## –11–

### 12 ● 

### Performance 

## Standardsfor the Food Stamp Employment and Training Program

Table 2—Comparison of Initial National Standards

Measure FNS OTA

Participation
rate. . . . . . . . . . .

Job placement rate. . . . . . . . . . .

Average wage. . . . .

Quarterly earnings. . . . . . . .

Food stamp case closures. . . . . . . .

Educational improvements. . . 10 percent

25 percent of terminees

$4.45/hour

20 percent of terminees

25 percent of those who start educational
programs Agrees—target more

intensive services on a smaller group

Agrees-based on FNS analysis of what is readily achievable

-$1,39 5

Would not use this

measure

Agrees-little data to set a standard

SOURCE: OTA, 1992.



## issues.” These goals are: 1) greater State flexibility, to accommodate the act’s call for greater services to participants with greater barriers to employment and for more emphasis on educational improvement; 2) establishing priority of service, with hard-to-employmandatory participants receiving first priority, mandatory participants second priority, and volunteers third priority; and 3) making use of existing data on FSET and its participants, specifically the Abt evaluation. 39



## OTA derived four criteria from its analysis ofcongressional intent. First, OTA suggests that setting and implementing performance standards be viewed as an ongoing 

## process, rather than a one-time activity. Congress recognized that modifications would be needed as State experience in operating FSET grew, and called for such modifications inboth the Food Security Act and the Hunger Prevention Act. In addition, OTA supports FNS’ proposal that ongoing modifications be used to encourage the States to continually improve on their previous performance. Second, OTA suggests that the performance standards be aimed at the goal, stated in the Hunger Prevention Act, of helping food stamp recipients 

## achieve self-sufficiency. Third, the standards would ideally be set to allow maximum 

## State flexibility, reflecting the Food Security Act which directs the States to implement a program, “designed by the State,” and including one or more of a list of five components. 40 OTA's 

## fourth criterion is that the model standards should encourage service to 

## individuals who have greater barriers to employment, as required by the Hunger Prevention Act.



# FNS AND OTA STANDARDS

Based on the criteria just outlined, OTA agrees with three of the five performan ce measures proposed by FNS. First, although it is not an outcomebased measure, OTA supports FNS’ decision to continue to measure State participation rates, and to require a minimal level of participation. OTA also supports FNS’ selection of two other outcome-based

measures:

. job placements; and

● 

## educational improvements.

However, OTA proposes to measure earnings, rather than wages, and OTA disagrees with FNS’ proposal for measuring reductions in food stamp dependency. The rationale is discussed below, along with a proposed standard for each measure and a method of calculating overall State performance.



## Participation Rate

On average, the States are meeting the current participation rate requirements. For example, in FY 1989, when the required rate was 35 percent, the States either enrolled or initiated sanctions against approximately 47 percent of non-exempt mandatory work registrants, or 1.2 million persons. 41 

## However,given the Abt evaluation showing that broad distribution of very inexpensive employment and training

3956 Federal Register, 43157 (Aug. 30, 1991).

4oJ0b ~em~ job ~~~~h ~~g; ~orkfme; work experience; ~ppofied work. T& Hunger Revention Act of 1$)88 added a Sixth COmponent+ducationrd programs. 41- et ~,, op. cit., fOOtnOte 15$ 

### P. 3-26”

### Section 4--Comparison Between FNS' Proposed Standards and OTA’s Model Standards . 

## 13

services 42 

## to a large fraction of work registrants had no impact, OTA supports FNS’ proposal to lower the required participation rate standard to 10 percent. As discussed below, OTA uses a different definition of “participant” than that proposed by FNS.

When reauthorizing funds for FSET in 1990, Congress limited the basic State grant total to $75 million for FY 1991-95. With this funding cap, only a lower participation rate will allow the States to provide more intensive employment and training services. To ensure that this smaller group (10 percent) of FSET participants includes those “with greater barriers to employment” (defined by FNS and OTA as those who lack a high school diploma and who had no work experience in the year prior to entering FSET), OTA agrees with FNS’ proposal to give extra credit in calculating some of the measures for those participants who fall into this category .43

In addition to Abt’s findings that FSET spreadslow-cost services too widely, research by the Manpower Demonstration Research Corp. (MDRC) suggests that programs that include at least some intensive education and training services have a greater impact on participants’ employment and earnings. The programs that were closest to FSET ingoals were the Work Incentive (WIN) Demonstration programs aimed at AFDC recipients and operated by the States with Federal assistance under the Omnibus Budget Reconciliation Act of 1981 (Public Law 97-35).

The WIN demonstrations served a different clientele from that served by FSET. WIN participants were overwhelmingly single mothers while FSETparticipants in FY 1988 were half male and predominantly childless. 44 

## Because of this difference, care must be taken in using research on the WINprograms to draw conclusions about ways to improve FSET. Nevertheless, there are striking similarities. The WIN programs resembled FSET in that most were relatively simple, low-cost programsemphasizing job search rather than extensive remePhoto credit: U.S. Department of Agriculture

A job search training class operated by the Wilmington, Delaware FSET program in 1988. Funding limits and a high participation standard meant that, nationally, less than one-third of FSET participants received

such training in FY 1988.



## diation or occupational training and in the fact that they were mandatory. As in FSET, the welfare agencies operating the WIN programs tried to enroll a large fraction of mandatory participants; however, another similarity with FSET was that the welfare agencies implementing the programs were reluctantto sanction those who failed to enroll or subsequently dropped out, and only about half of the ‘‘mandatory’ participants ever actually participated for even a brief time. .45



## Despite these similarities, FSET had no discernible impact in 1988, while five of seven WINprograms that MDRC evaluated using an experimental design yielded statistically significant gains in earnings for participants. 46 

## The average earnings among the experiment groups in the WIN programs ranged from 10 to 30 percent more than the earnings of the control groups by the third year after program entry .47 There are two key differences between WIN and FSET that may account for this. First, although low in cost, the WIN programs were more expensive (ranging from under $200 per participant in Cook

42Just overhlfof RN3’T clients were assigned to individual job seareh 27 percent to job se=ch training, 6 percent to work experience, and 16 percent to educational services in FY 1988. w et al., op. cit., footnote 15, p. 6-16.

43~s proposes @v~g exha cr~it o~y in the job p~cement and wage measures. OTA suggests giving exm Crdit not o~Y in ~e~ two m~~s~ but also in the educational improvement measure. This is discussed further below.
44- et ~., op. cit., fOOtnOte 15, P. 3-30.

45Judith M. Gueron, “Work and Welfare: Lessons on hnployment fiogrms, ” Journal of Econom”c Perspectives, vol. 4, No. 1, winter 1990, p. 90. ‘Ibid., p. 87.
d7~e exceptions were west v~~a, a s~~ ~~ exceptio~y hi@ ~~ploma~ ~d Cwk Co@y, I~inois, which operated the l-t expensive

prograq monitoring individual job searches and providing little direet assistance. Ibid., p. 94.

### 14 ● 

### Performance 

## Standards for the Food Stamp 

### Employment and Training program



## County, Illinois and Arkansas to around $1,000 in Baltimore) and involved more of participants’ timethan FSET. Second, as discussed above, the population served by the WIN programs was made up of mothers, with little job experience, while that served by FSET is half male and predominantly childless.

Support for the hypothesis that FSET had little impact because its per participant expenditures were so low is found among the seven WIN programs. As shown in table 3, the only programs that had no impact were those in West Virginia, which had the highest unemployment rate in the Nation (21 percent in 1983), and Cook County, Illinois, where average expenditures were only about $130 to $160 per participant. 48 

## The low costs of the Cook County program reflected its emphasis-monitoring and sanctioning nonparticipants. Little staff time or expense was required for direct services, because the program consisted of 2 months of independent job search followed by up to 3 months of unpaid workfare. 49 

## The Arkansas program, while also inexpensive, provided more direct assistance to participants, primarily in the form of a 2-week job club. For comparison with FSET, see footnote 42.

By contrast, the programs that spent the most on their participants, in Maryland and California, yielded the largest absolute gains in participant earnings when compared with control groups. San Diego’s Saturation Work Initiative Model (SWIM) program, which combined an initial period of job search for all participants followed by unpaid work experience, remediation, and occupational training for those who needed it, yielded the greatest earnin gs gains, which increased to nearly $900 per year by the third year after participants entered the program. 5o 

## In its most recent analysis of the seven WIN programs aswell as several others, MDRC concluded, “broadcoverage programs that included some higher-cost services had greater average e arnings impacts than those that did not.’ ’ 51 

### Table 3-Costs and Earnings Impacts of Mandatory Employment and Training Programs

for AFDC Recipients

Gains Gains Program costs (dollars) (percent)

Illinois (Cook County). . . . . . . . . . $157 $0 0%

Arkansas. . . . . . . . . . . . . . . 118 337 31 (year 3)

West Virginia . . . . . . . . . . . . 260 0 0

Virginia. . . . . . . . . . . . . . . . 430 268 11 (year 3)

California (l). . . . . . . . . . . . 636 443 23 (year 1 )

California (II) (San Diego SWIM). . . . . . . 919 658 29 (year 2)

Maryland . . . . . . . . . . . . . . . 953 511 17 (year 3)

NOTE: Costs are average cost per participant. Gains are statistically significant increases in the average amount earned by participants when compared with controls.

SOURCE: Judith M. Gueron and Edward Pauly, From We/fare to i4brk

(New York, NY: The Russell Sage Foundation, 1991), pp. 15-17.

Why 

## would the second difference, a higherproportion of males with fewer dependents, contribute to FSET being ineffective? One possibility is that single mothers have so little labor market experience that even small interventions will give them a significant gain in employment and earnings, whereas males and single females who are the targets of FSET generally have some job experience and many of them can find employment without the help of an employment and training program. 52 

## However, the results of the San Diego SWIM program suggest that more intensive services 

## can have an impact, helping males not only to find jobs, but to find better jobs than they otherwise would have. This program included males receiving AFDC-UP (unemployed parent) benefits. On average, male participants earned $454, or 12 percent, more than males in the control group in the second year after random assignment. 53



## Another possibility is that mothers are more likely to benefit from employment and training programs

481986 d~llars.

4~~el F~~~&~ et & Fi~l Report on Job search And workExperienCe in cook Counfy mew York NY: Wpower Demo n Corp., 1987), p. viii.
5oJudith M+ Gueroq cited fi J.f~npoWerDe~m@~tion R~S~~rch corporation: lf)ag Annual Report (New York NY: M

Sljudith M. Gueron ~d ~w~d pa~y, From we~are to work @Jew yor~ NY: Russell sage FouditioQ 1991), p. 27

. 52~ong arepresen~tive smple of lq,~workregis~ts eligible for FSET in 1988, about half(53 per~nt) bd hd some employment h year. The 13,000 work registrants were randomly assigned to participate inFSET or enter a control group. When participants and controls were surveyed 1 year after random assignmen~ about half (52 percent) of each group had experienced some employment during the previous year. Run% et al., op. cit., footnote 15, pp. 3-31 and 4-25. sqGueron and Pauly, op. cit., footnote 51, p. 35.

### Section 4--Comparison Between FNS’ Proposed Standards and OTA’s Model Standards ● 

### 15



## because they are highly motivated by the desire to provide for their children. Both the mandatory WIN programs and voluntary demonstration programs, including one which trained AFDC recipients as home health aides, 54 

## have increased the earnings of single mothers. The literature does not suggest whatmight provide similar motivation to males or females without children. However, one approach to increasing the motivation of FSET participants might be to make the program voluntary, rather than mandatory (see section 5). It is possible that those who are self-motivated to volunteer would be more likely to complete education, tr aining, job search, and other activities that would lead to employment than those who are motivated only by the threat of sanctions.

Definition of “Participant”
FNS proposes to use two different definitions of “participant.’ For the purposes of meeting the 10 percent participation standard, a “participant” would be any non-exempt mandatory work registrant who is assigned to FSET, including those who never appear for an FSET activity .55 For the purposes ofcalculating the outcome-based performance standards, FNS proposes to use a base of ‘‘terminees’ made up of only those participants who actually begin their frost assigned activity.

OTA finds these two definitions inconsistent and suggests that FNS use a single definition for both purposes-those who actually begin their first assigned activity, excluding initial “no shows. ” Interms of the participation standard, the Abt evaluation discussed above found that one factor in FSET’s lack of impact was lack of participation. Despite the threat of sanctions, 56 

## over half of “mandatory” participants never received any employment and training services that year. Most of these (34 percent) were ‘‘no shows,” who never appeared for their frost FSET activity, such as a job club. Thus, to meet the 10 percent participation standard under OTA’s definition, the States might have to target about 14 percent of work registrants for enrollment with the expectation that one-third would not appear.

OTA agrees with FNS’ proposal to include volunteers along with mandatory participants whencalculating the outcome-based performance measures.

Priority of Service

FNS proposes that, in delivering FSET services,the States give first priority to mandatory participants (i.e., work registrants who have not been found exempt) who are “hard to employ” (HTE), second priority to mandatory participants who are not hard to employ (NHTE) and third priority to food stamp recipients who volunteer. 57 

## As an incentive to the States to implement these priorities, FNS proposes giving extra credit for HTE in the job placement andwage measures, regular credit for mandatory participants in all four measures, and half credit for volunteers in the job placement and educational improvement measures. 58



## Based on the criterion that FSET should serve 

## those with greater barriers to employment and onMDRC’s research which shows that welfare-towork programs have little impact on those who are most job-ready, OTA supports giving first priority to HTE participants. In the WIN programs discussed above, program impacts were greater-up to a point-for those who were more welfare-dependent and less job-ready .59 However, the qualification is significant as figure 4 shows: impacts were smallest for the most 

## welfare-dependent as well as the 

## least welfare-dependent (who were most able to find jobs on their own). 60 

## In addition, the positive impactsoccurred in settings where the more welfaredependent were “mainstreamed” in job clubs with those who were more job-ready, a setting which may have benefited the less job-ready. This research implies that FSET might begin to have an impact if

fistephenH. Bell et~o, Ovemiew ofEvaluationResults-Eval~tion 

### oftheAFDCHomemaker-HomeHealth AideDe~mt@iOm (w-o%~:

Abt Associates, Inc., 1987), p. ix.

5556 Federal Register 43188 (Aug. 30, 1991).

S6Abt fomd tit ~actiom ~me not s~ongly ~ppli~. However, o“l’A’s ~ysis, &~sed ~a in Sation 5, suggests th sanctions do little to encourage participation in “mandatory” programs.

5756 Federal Register 43157 (Aug. 30, 1991).

58~s @vw

~ ~r~t

for ~olut~rs in its ~~e ra~ ad food stamp
~ clos~e m-es. The ratio~e for this inconsistency k tlllcle~.

5~~el Ffiedlmdw ad David ~%, A Sttiy of pe@or~nce Measures ad subgroup rmpacts in Three we~are Emp l
(Washingto~ DC: National Commission for Brnployment Policy, 1987), pp. vii, xv.
‘Ibid., p. x.

### 16 ● 

### Performance 

## Standards for the Food Stamp Employment and Training Program

Figure 4-Subgroup Impacts



# r

$!

w 

# H

o I I 1A

..%%%.

. . 

# 1,

No Prior AFDC AFDC returnees Long AFDC or no earnings

Increasing welfare dependency --> 

# ❑ Baltimore 

# ❑ Virginia 

# ■ Arkansas

The impact of employment and training programs is greatest for those at an intermediate level of welfare dependency. SOURCE: Daniel Friedlander, Performance /rrdicators for Se/ecfed We/fare Employment Programs (New York, NY:

Manpower Demonstration Research Corp., 1988), p. xiv.



## the States could be encouraged to move beyond the most employable (the least welfare-dependent) to serve the middle group. It also suggests that very intensive services are needed to have an impact onwork registrants in the bottom (most welfaredependent) group.

OTA concurs with FNS’ definition of HTE-- those who have not completed high school or obtained an equivalency degree and have not been employed in the preceding 12 months. However, based on its criterion that FSET should be aimed at helping individuals achieve self-sufficiency, 

## OTA proposes that full credit be awarded for successful outcomes among volunteers.
Many work registrants leave the food stamp rolls and find work (with or without receiving FSET services), only to become unemployed and go back to needing food stamps again. 61 

## To achieve selfsufficiency, this group may require remedial education and/or occupational training” to obtain more stable employment. Although there is no conclusive research on this point, volunteers might be more likely than mandatory participants to have the self-motivation required to complete such training and to search for a good job. While allowing full credit for volunteers, OTA recognizes that this could lead to providing services to the most job-ready, who are least likely to benefit from FSET. Therefore,OTA suggests giving extra credit for HTE volunteers as well as HTE mandatory participants incalculating the other three outcome-based performance measures.



## Job Placements
FNS proposes the following measure of entered employment rate (EER):
EER = (3 X no. HTE mandatory jobs) + no. NHTE mandatory jobs + (0.5 X no. volunteer jobs)/HTE

61c~le~ U&ret ~o, ‘‘Long ‘Mm Participation in the Food Stamp Program by Work Registrants, “ in Carole Trippe et al. (eds.), Food Stamp Policy
Issues: Results From Recent Research (Alexandri% VA: U.S. Department of Agriculture, Food and Nutrition Service, 1990), p. 205.

### Section 4--Comparison Between FNS’ Proposed Standards and OTA’s Model Standards ● 

### 17

Table 4-Job Placement Rate-Comparison Between FNS’ Proposed Rule and OTA’s Model Standards

FNS OTA



### . . . . . — Definition. . . . . . . . Participants employed within
1 month after the month they leave FSET

Weights for HTE. . . . . . . . . . Triple credit for

HTE, based onrelative employability of the two groups

Priority of service. . . . . . . Half credit for

volunteers

Weights for benefits. . . . . . . No extra credit

for jobs with medical benefits Participants employed



### at 3 months after leaving FSET

Same-agree with analysis and logic. Include HTE volunteers

Full credit for volunteers, who may be more motivated than mandatory participants

Double credit for jobs with medical b benefits, to increase mtion

SOURCE: OTA, 1992.



## mandatory + NHTE mandatory + volunteer terminees.

OTA agrees with FNS that State performance should be measured, at least in part, in terms of job placements. Two of the goals of FSET are to increase employment and to help individuals achieve self-sufficiency. However, as noted above, OTA would give full credit to volunteers (and extra credit to HTE volunteers) in calculating this and the other performance measures. OTA agrees with FNS’ proposal to include volunteers in both the numerator and the denominator for this measure.

OTA disagrees slightly with FNS on the timeframe for measuring employment outcomes of FSET participation. FNS proposes “the end of the month following the month of E&T (employment and training) termination.” 62 

## This would be between 1 and 2 months after completion of FSET activities. Because job-finding can be a drawn-out process, especially when the economy is weak, OTA suggests an alternative measuring period corresponding with that used under JTPA performance standards— 13 weeks after leaving the program (see table 4). 63 

## OTA suggests this timeframe not only to conform to JTPA, but also because employment status at 13 weeks (or 3 months) from program exit is a betterpredictor of long-run gains in earnings and reductions in welfare than employment status in the month following program exit. 64



## FNS does not specify how data for this measure are to be obtained, but appears to assume that the States will conduct postprogram surveys and/or interviews with a sample of participants. OTA agrees with this approach, but suggests that FNScontinue to monitor the feasibility of using unemployment insurance (UI) wage record data. When current problems of interstate data transfer and time delays can be resolved, UI data may prove to be moreaccurate and cost-effective than relying on participant surveys. 65



## Weights

OTA supports FNS’ proposal that job placements be weighted to encourage the States to enroll those ‘‘with greater barriers to employment. The Hunger Prevention Act requires the States to help this group; in addition, such targeting may represent the best use
of limited Federal and State FSET funds.
Without weights, and with a participation standard of only 10 percent, the States might simply enroll the most job-ready work registrants to meet the job placement standard. FNS found that, among 1988 FSET participants, the HTE group was only half as likely to find jobs as the NHTE group. 66 

## Recognizing that it might take the States at least twice as much effort to place this group, and to provide an additional incentive, FNS proposes giving triple credit for each HTE placement. OTA concurs with FNS’ analysis and proposed weighting.
In order to move toward the goal of helping ‘‘individuals achieve self-sufficiency,’ OTA alsosuggests that additional weight be given for place6Z56 Federal Register 43158 (Aug. 30, 1991).

63Ro&~ ~ Jonm, ~~TrU ~d fiplommt Gfi~ce ~tter No. $L8$I, J~e 29, 1$YJ(),” U.S. Dep@ent of bbr mimeo, Washhlgto% DC.
64Jefiey ~fi~~ et & E~tabli~hing

~ Pe@oMnce Ma~gement Sy~temfor Targeted Weyare program (w~~to~ DC: National cOK13UlkSiOIl for Employment Policy, 1988), p. vi.
&Jo~ BaJ ~d Chles E. Tm~ A Fea3ibili~ s~dy of th

e use of Unemployment In~rance wage-RecordData as an Evaluation Tool for JTPA (Washington DC: National Commission for Employment Policy, 1991), p. v.

6656 Federal Register 

### (Aug. 30, 1991), p. 43168.

### 18 ● 

### Performance 

## Standards for the Food Stamp Employment and Training Program



## ments of both HTE and NHTE clients (includingvolunteers) in jobs that include paid medical benefits. According to a recent study, “low-wage jobsgenerally come with low-or no-medical benefits.” 67 

## This problem prevents some welfare recipients, who often qualify for Medicaid, from leaving the welfare rolls. Congress recognized this problemin providing transitional Medicaid for JOBS participants who obtain employment, but no such benefits are available for FSET participants who become employed. Based on a 1986 estimate that only 13.4 percent of low-wage jobs provided health benefits, it maybe seven or eight times as difficult to find such

jobs. 68 Following FNS’ 

## approach to weighting for HTE job. placements would lead OTA to propose seven or eight times credit for placements in jobs with paid medical benefits. However, to avoid placing too much emphasis on this one measure of job quality, OTA suggests giving double credit for jobs with medical benefits. This extra credit would apply to all FSET participants, whether or not they are hard to employ, including volunteers.

Initial National Standard
Based on its analysis of State success rates in placing FSET participants in 1988, FNS proposes an initial benchmark of 25 percent for FY 1992-94. This proposed standard is based on what FNS estimates that 75 percent of the States were able to achieve in FY 1988. 69



## FNS proposes raising the standard in future years, as the States become better at finding jobs for FSETparticipants. For example, a new national benchmark for FY 1995-96 would be set based on State perfo rmance during FY 1992-94. However, futurestandards, like the initial benchmark, would continue to be based on the 25th percentile of what has been achieved. This proposal follows Congress’ intent that the new performance standards be easily attainable. OTA concurs with this approach and with its rationale. OTA notes that the use of the 25thpercentile does not preclude increases in the standard over time. Using the 25th percentile, the JTPA Photo credit: U.S. Department of Agriwlture

The 25 percent placement standard reflects the difficulty of placing even job-ready FSET participants. This woman with job-related skills went through four tough interviews with the help of an FSET counselor before landing a job as a clerical worker in an appliance store.



## performance standards for adult followup employment increased from 60 percent for program years 1988 and 1989 70 

## to 62 percent for program years 1990 and 1991.



## Earnings

The Hunger Prevention Act specifies that the FSET performance standards take into account job placement rates, wage rates, and job retention rates. However, FNS’ proposed standards take only the first two factors into account. FNS proposes the following measure of wage rates:
Average hourly wage rate = (0.5 X portion exceeding minimum wage) + HTE wages + NHTE mandatory wages + volunteer wages/HTE + NHTE.

OTA suggests that FNS develop an alternative

measure--e arnings--that would capture both wage levels and job retention. Most FSET participants in 1988 were very poor. Although 53 percent had experienced some employment in the year prior to being certified for food stamps, 71 nearly three



## quarters had annual incomes below 75 percent of the poverty line of $7,500, as defined by the Office of

‘%Iary Jo Bane and David T Ellwood, ‘“ Is American Business Working for the Poor?” Harvard Business Review, September-October 1991, p. 60. 6S0~y 13.4 Pmcent of persons ~~g 1ms

~ $5,~ fi 1986 remiv~ employer-provided he~th iIISUHUMX, Rob@ L. clm~ Employee Benefits for American Workers (WashingtorL DC: National Commission for Employment Policy, 1990), p. 19. In 1988, two-thirds of FSET participants earned less than $3,000. Puma, et al,, op. cit., foomote 15, p. 3-31.

6956 Federal Register 43187 (Aug. 30, 1991).

TO~A opmtes on a 

## progr~ year, from July 1 to June 30, rather than the Federal fisd yem. running from Oct. 1 to Sept. 30.

71- et al., op. cit., footnote 15, p. 3-31.

### Section 4--Comparison Between FNS’ Proposed Standards and OTA’s Model Standards . 19



## Management and Budget. Eighty percent earned less than $6,000 in the year before they entered the

program. One factor in the poverty of the work registrants targeted for FSET is the short duration of the jobs they find. An analysis of work registrants in the States of Alabama and Washington found that thelargest share of food stamp allotments “was consumed by households that experienced multiple spells within 2 to 21/2 years 

## after they were initially certified. ‘’ 72 

## Although they made up less than a third of the work registrants studied, these households consumed nearly half of all food stamp allotments during the study period.
OTA suggests a measure of earnings, not only because earnings reflect job retention, but also because earnin gs may ill uminate the impact of the program. In a study for the National Commission forEmployment Policy, Abt Associates examined several alternative short-term performan ce measures to see how well they correlated with real earnin gs gains (identified in a comparison with control groups) for AFDC recipients who had been involved in a demonstration project to employ them as home health aides. Abt found that average earnings over the 6 months following program termination was the best predictor of long-term gains in income and reductions in welfare dependency. 73 

## Given the high cost of obtaining data at 6 months after program exit, Abt found that average wages over 3 months was nearly as good a proxy for earnings gains.
OTA proposes that earnin gs be measured inconjunction with employment, either through interviews with participants, or, if feasible, using UI wage data on additional data tain. 74



## Weights participants. OTA finds that the

would not be burdensome to obFNS proposes that weights be used to encourage the States to place FSET participants, particularly those who are hard to employ, in jobs paying above the minimum wage. Specifically, FNS proposes that the States be given 1.5 times the usual credit for that portion of the wages of HTE participants that exceeds the Federal minimum wage ($4.25 per hour, as of April 1, 1991). For example, an HTE client who was placed in a $6 per hour job would be credited at an average wage of $6.87 per hour. FNS’ proposal, like that for job placement of HTE, is based on itsanalysis of the wages earned by HTE FSET participants in 1988. This analysis found that the average wages of HTE participants were above the minimum wage, but by a smaller amount than the wages of NHTE participants. 75 

## OTA finds the analysis and proposed weighting reasonable, and suggests that FNS incorporate them into an earnings measure.

In assessing overall State performan ce, OTA suggests that extra credit be given for achievement of the earnings standard. A study of the impacts of performance standards on the services provided by local JTPA operators found that policies which put greater weight on the wage standard “significantlyincrease provision of classroom training in occupational skills. ’ ’ 76



## Initial National Standard

FNS proposes an initial national wage benchmark of $4.45 per hour, or 5 percent above the national minimum wage that became effective April 1, 1991. As with the other initial standards, the $4.45 level is derived from its analysis of data derived from the 1988 FSET evaluation, and is expected to be the level that at least 75 percent of the States are already achieving.

OTA suggests an initial quarterly earnings benchmark, to be measured at 3 months following program exit, in the range of $1,395 (see table 5). This estimate is based on the average number of hours worked by 1988 FSET participants, multiplied by a

TzCharles Usheret al., “Ixmg-’IkrmParticipation in the Food Stamp Program by Work Registrants, “in Carole Trippe et al. (eds.), Food Stanzp Po2icy
Zssues: Results From Recent Research (Alexandria VA: U.S. Department of Agriculture, Food and Nutrition Service, 1990), p. 205. %ZOrXIitSky,

et al., op. cit., footnote 64, p. vi.

74~e u-s. D~~~nt of ~~r, ~h,ich Ovasees co~wtion of posQmW~ &@ on JTPA p~cip~ts,

~ fowd tit the greatest ~~lEw iS fhd~ the persons to be interviewed. Adding afewmore questions to the interview does not add greatly to the time or expense of data gathering. Steve AaronsoQ U.S. Department of Labor, personal communication Nov. 27, 1991.

7556 Federal Register 43171 (Aug. 30, 1991).

TGDic~on et ~., op. cit., footrlOte 39, p. 195.

### 20 . Performance Standards for the Food Stamp Employment and Training 

## Program

Table 5-Average Wage and Earnings-Comparison Between FNS’ Proposed Rule and OTA’s Model Standards

FNS--Average wage OTA--Quarterly earnings

Definition. . . . . . . . . . . . . . . . . Average wage of those em- Earnings in the 3 months after ployed within 1 month of leaving FSET leaving FSET

Weights—HTE. . . . . . . . . . . . . Extra credit for HTE wages lncorporate extra weight into above $4.25/hr (the current min- earnings measure, with extra imum wage) credit for HTE wages above $5.00/hr.

Weights--overall performance. .



### . . . . . . . . . . . No extra credit for wages in Extra credit for earnings in asoverall performance sessing overall State performance--to encourage placement in longer lasting jobs

SOURCE: OTA, 1992.



## wage of $5 per hour. 77 

## FNS may wish to calculate a more exact earnings benchmark, based on the 25th percentile, rather than the average.
OTA’s proposed earnings standard incorporates a slightly higher wage ($5 per hour) in order to encourage bringing more FSET participants out of poverty. The Abt evaluation found that most FSET participants lived in relatively small households; nevertheless, the average household size was 2.2 persons, and 30 percent lived in households with three or more children. The Bureau of the Census places the 1990 poverty line for a family of three at $10,419. An FSET participant who was the sole support of such a household and was placed in a job earning the benchmark wage proposed by FNS ($4.45 per hour) would earn only $9,256 under the optimistic assumption that the job was full time and lasted a full year. Thus, the family would still be in poverty. Only single and two-person food stamp households would be brought above the 1990 poverty line with the wage proposed by FNS. Byincreasing the benchmark to $5, FNS could encourage the States to seek out better jobs that would represent a significant improvement over welfare.
An earnings measure based on a $5 per hour wage should not be too difficult to attain. Among food stamp recipients enrolled in JTPA in 1989, the average hourly wage at termination was $5.11, but the average hourly wage among those who had participated in classroom training was $5.56, while those who received only job search assistance averaged $4.75. 78 

## Another study of JTPA participants in 1985 found that training correlated directly with wages—training in higher skill jobs led to higher average wages for both those who were more job-ready and those who were less job-ready (this group included welfare recipients with no recent work experience). 79



## Reductions in Food Stamp Dependency

FNS proposes the following measure of reductions in food stamp dependency:
Case closure rate = no. all mandatory cases closed + no. all voluntary cases closed/all mandatory terminees and voluntary terminees.
Although the FNS proposal is derived from the Hunger Prevention Act, which specifies that the performan ce standards take into account “households ceasing to need benefits under this act,” OTA disagrees with it. OTA 

## does agree with FNS’ statement that “it would be extremely difficult to construct an affordable measure that could directly capture ‘ceasing to need benefits.'"80Need for food

77~~ ~v~.~g~ hOm~ ~~&.d fi the 90 &y~ fo~o~g Progm ~nt~ w~ 279. ~o~tion 

## Suppfied by Steven Bell, Abt Assochtes, hc., Wd on Puma et al., op. cit., footnote 15, 

### pp. 7-15,7-19.
Tgu.s+ 

## Dep~ent of Labor, op. cit., footnote 18, 1991, p. 29.

v~owever, the study also found that less jobready individuals typically 

### reeeived less training and more low-cost job seareh assistance than those who weremorejobready. In addition, dropouts were undeserved in proportion to the eligible population, and typically received little remedial edueation and little training. U.S. General Accounting OffIce, Job Training Partnership Act: Services and Outcomes for Participants With Di#en”ng Needs, GAO/HRD-89-52 (Washington, DC: 1989), 

### pp. 3,51.

~56 Federa/Register 43172 (Aug. 30, 1991).

### Section 4--Comparison Between FNS' Proposed Standards and OTA’s Model Standards .21



## stamps is a complex matter determined on the basis of family size, income, living expenses, and other factors by caseworkers on a case-by-case basis. Obtainin g this information from individuals who are no longer involved in FSET and possibly no longerobtaining food stamps would be costly and timeconsuming.
Although FNS’ proposed substitute is a much simpler measure, case closures can occur for manyother reasons besides increased income. In particular, using case closures as a performance measure could have the unintended effect of encouragingStates to sanction nonparticipants, rather than encouraging participation through positive means.
MDRC’s studies of the mandatory WIN programs found that, among the most disadvantaged subgroup participating, there were no “consistent or large” earnings gains, but that this group provided the largest welfare savings. 81 

## This suggests that some of those eligible for AFDC went off the rolls, perhaps out of fear of sanctions, without finding jobs. Using food stamp case closures as a performance measure could have the same effect on the HTE group within FSET, discouraging those who most need food stamps from using them.
Although omitting this measure does not accord with the original language of the 1988 Hunger Prevention Act, Congress recently changed its requirements for the FSET performance standards 82 

## and no longer specifies that they take into account ‘‘households ceasing to need benefits.’



## Educational Improvements

. . . today, even if you’re a janitor, there’s still reading and writing involved. Like if they leave a 

## note saying, ‘Go to room so-and-so, this and that. ’ You can’t do it. You can’t read it. You don’t know. And they ain’t going to hire somebody to run along and tell people what to do. 83 

## FNS Proposed Measure

FNS proposes the following measure of educational improvement:

Educational improvement rate = no. mandatoryterminees + (voluntary terminees x 0.5) who completed 64 hours, or earned General Educational Development certificates (GEDs), or met State goals/ all educational mandatory and voluntary terminces. w



## This calculation measures the proportion of ‘successful’ participants among those who enter educational programs. Successful completion would be determin ed by the States, 85 

## judging performance on one or more of the following:
1. completion of 64 or more hours of classroom training; 2. obtaining a high school diploma or equivalent;

or3. meeting State-defined educational goals comparable to those that could be accomplished in the 64 hours prescribed in 1. 86



## Education that can be counted toward the 64 hours must focus on ‘‘basic educational skills. ” 87 

## Examples are adult basic education (ABE) and english as a second language (ESL) classes, GED preparatory classes, and other training designed to improve the skills of participants in one or more of the six basic skills of reading, writing, mathematics, speaking, listening, and problem-solving. 88 

## FNS proposes 

## that inclusion of an educational improvements measure be at the option of each State. No explicit penalty would be levied on States that did not include such a measure. However, successful performance on the measure would earn States extra credit toward receipt of incentive funds. 89



## This section first considers the justification for including educational improvements as one of theoutcome-based performance measures. It then pre81 Gueron ~d Pauly, op. Cit., fOOhIOk? 51, P. 30.

82The Ford, A@-~@.e, Comenation ~d ‘f’~ade Act of 1991 @blic ~W 102-237, signed December 13, 1991).

83EllenCole, TheExpenenceoflz/iteracy ~e~owspfigs, oH: Ufion ~ad~te school, l$)7@, 

## p. 123 i,u paulaDugg~Literacyat Work; Developing

Adult Basic Skills for Employment (Washingto~ DC: Northeast-Midwest Institute: The Center for Regional Policy, 1985), p. 2.

~56 Federal Register 43173 (Aug. 30, 1991).

8556 Federal Register 43159 (Aug. 30, 1991).

8656 Federa/Register 43172 (Aug. 30, 1991).
gvIbid.

8856 Federal Register 43172 (Aug. 30, 1991). rbid.

8956 Federa/Register 43184 (Aug. 30, 1991).

### 22 ● 

### Performance 

## Standardsfor the Food Stamp Employment and Training Program



## sents an analysis of the FNS proposed measure, and OTA’s suggested changes.

Justification for Measurement of Educational Improvements
Although States are not required to raise the educational level of FSET participants, there are good reasons to encourage them to do so byincluding educational improvements in the outcomebased performance measures. Arguments in supportof this position include legislative intent, the relationships between education, employment and earnings, and employer demand for higher educational levels in applicants for positions where demand for educational qualifications has been, until quite recently, minimal.
Congressional interest in education as a means toreduce food stamp dependency has increased steadily since the mid-1980s. In the Food Security Act of 1985 Congress defined a food stamp employment and tr aining program as including one or more of thefollowing components: job search, job search training, workfare, and work experience and/or training. 90 

## In the 1988 Hunger Prevention Act, Congressadded “educational programs or activities to improve basic skills” to this list of optional components. 91 

## The Food, Agriculture, Conservation, and Trade Act of 1990 inserts “and literacy” after “basic skills” in the phrase just quoted from the 1988 act. 92



## There are strong associations between education, employment, and earnings. The more education a person has, the more likely that person is to be employed. Among those employed, the more highly educated enjoy higher wages, 93 

## and over the past 30 years, education has increased in importance as a contributor to earnings. Between 1960 and 1984, the difference in earnin gs between high school graduates and those who dropped out of high school increased from 30 to 60 percent. 94 

## The positive associations between education and employment and between education and earnin gs are likely to increase in the future, paralleling expected increases in both the pace of change and the complexity of technology.

Employers increasingly use educational attainment as a screen in both hiring and promotions. A postsecondary education is now desired in many industries where job applicants could previously qualify with a high school diploma or less. For example, textile firms now seek graduates of 2-year associate degree programs for maintenance and repair positions, a striking change from past practice. Until recently, textile companies saw no need for a high school diploma, and entry-level workers with motivation and ability could expect to rise fromlow-level jobs into repair and supervisory positions. 95 

## On average, in all industries, workers with less formal schooling have fewer opportunities forworkplace-based training, and thus for advancement, than those who are more highly educated. 96



## Compared to the general population, the food stamp population is poorly educated. Whereas three-fourths of the U.S. population aged 25 and older have completed a high school education, 97 

## half of those eligible for FSET had done so in FY 1988. 98 

## In a 1986 study of food stamp work registrants in job search demonstrations, average schooling was 10.5
years. 99 

## FSET participants are 

## even less likely than

the 

## total population of food stamp recipients to have a post-high-school degree. loo

%e Food Security Act of 1985 (Public Law 99-198), Title X, Subtitle

~ sec. 1517. gl~e H~ger prevention Act of 1988 (Public Law 100-435), Title 

### IV, SW. 404.

~~e Food, Agriculture, Conservatio~ and Trade Act of 1990 (Public Law 101-624), sec. 1726. gsRes~ch on this point iS S ummarized in Gordon Berlin and Andrew Sum, Toward a More Perfect Union: Basic Skills, Poor Fam”lies, and Our Econom”c Future (New Yorlq NY: Ford FoundatiorL 1988), p. 40. 94~thony p. cmev~e, ~i~ J. Gfim, ~d Ann S. Mel~r, workplace Basics: The Essential SkiZ/s Employers Want (SaII Francisco, C A Jossey-Bass, 1990), p. 5. 9SOTA, Wortir Training, op. cit., footnote 4, pp. 84-85. $I150TA, Worker Training, op. cit., footnote 4 P . 228.

%’Michel

_ Mm

W

ern e r, ~d we Hoj~, EVal~fion of the Food stamp Employment and Training program: Repo r Program Zmp2ementation (Washingto~ DC: Abt Associates, Inc., 1988), p. 53.

98- et ~., op. cit., foo~ote 15, Evaluation of t h

e FoOdstamp Employment and Training program, Jwe 1990, table 3.2 ,
%randeis University and Abt Associates cited in Abt Associates, Report to Congress, Dec. 16, 1988, p. 53.
lmIbid., p. 53.

### Section 4-Comparison Between FNS' Proposed Standards and OTA’s Model Standards ● 

### 23



## The educational deficits of FSET participants are paralleled by unemployment and low household income levels. Half of FSET participants reported working for pay during the previous 12 months 101 

## compared to three-quarters of the general population over the age of 16. 102 

## Annual income is less than $3,000 annually 

## for about two-thirds of FSET participants, and about 80 percent had annual incomes of less than $6,000. 103 

## Given the relationships between education, employment and earnings, lack of education appears to be a major reason that food stamp recipients need food stamps. The word ‘‘appears’ is deliberate. As discussed above, research indicates that in the population as a whole, higher educational levels are associated with higher earnings, but there is no definitive evidence that providing education to welfare recipients enables them to be more self-sufficient. Suggestive evidence comes from programs that include significant amounts of education and training and show sizable impacts on earnings, but these programs differ from less successful programs on other dimensions thaneducation. Thus, the causative importance of increased education to welfare recipients is unproven

. l04 

## The lack of data on the impact of provision of educational services to adult welfarerecipients has significantly tempered OTA’s conclusions about an educational improvement measure.

Analysis of the Proposed FNS Measure of Educational Improvements

OTA makes the following observations andcomments about the educational improvements measure proposed by FNS.
The measure is optional.--Three argumentssupport making measurement of educational improvements optional when measuring State performance. First, neither Congress nor FNS requires the States to include educational improvement components in their FSET programs. 105 

## Second, current data are inadequate to develop a baseline for anational standard against which educational improvement may be compared. 106 

## Third and most important is the above-mentioned lack of direct evidence showing that more education increases the employment and/or earnin gs of those receiving welfare benefits.

Because of these valid arguments, OTA concursin making measurement of educational improvements optional initially. However, research now underway may demonstrate the extent to which education enhances the self-sufficiency of those on welfare. 107 

## OTA suggests that transition to a mandatory measure occur within 5 years if the value of educational improvement to persons in or similar to the food stamp population is demonstrated

Weights for the hard to employ. 108 

## —In no t giving extra credit to the HTE, FNS’ proposed educational improvements measure differs from the entered employment rate and the achievement of wages above the minimum wage. The reason given for no extra credit for educational improvements by the hard to employ is that one of the barriers defining them is their lack of a high school diploma or a GED and thus States are already motivated to include them in educational activities. l09 

## However, the rest of the definition of the hard to employ is that these participants have not worked in the last 12 months.Under the proposed standard for entered employment rate, States receive triple credit for HTE participants who obtain jobs. Triple credit under half the definition of HTE and no credit under the other half of the definition is not consistent. OTA suggests that consistency be maintained and that triple credit be given in both instances.

l“llbid., p. 56.

102Ryscavage and Felb-HmW in Repoti to Congress. . . . Ibid., p. 56.

los~ et al, op. cit., footnote 15, p. 3-31.

104Gueron ~d pa~ey, Op. Cit., fOOtiOte 51, P. ‘.
1O$5G Federal Register 43173 (Aug. 30, 1991).
1065(j Federal Register 43172 (Aug. 30, 1991).

107~RC*~ s~dy of ~ifo~~s JOBS progr~, w~ch will produce in~rim findings in dle SllmlIl er of 1992, and the Rockefeller Foundation’s continued research into the impacts of the Minority Female Single Parent Demonstration Program will provide more information on the value of education in enhancing employment and earnings.

10s56 Federal Register 43173 (Aug. 30, 1991).

10956 Federal Register 43173 (Aug. 30, 1991).

### 24 ● 

### Performance Standards for the Food Stamp Employment 

## and Training program

Volunteers receive half credit. 110 

## -Consistent with its proposals for other measures, OTA proposesthat volunteers who satisfy the educational improvements measure be counted at full credit.

The denominator is restricted to those who start an educational component.--One consequence of limiting the denominator to those who start an educational component is that a State where a few people start and all of them successfully complete an education component would score higher by this measure than a State where many people start an education component but only half successfully complete it. OTA notes that this problem could beavoided by changing the denominator for the educational improvements measure to all mandatory and voluntary participants. The larger denominator would imply that all those in FSET should participate in educational components, which seems inappropriate for an optional standard. Therefore, OTA concurs with restricting the denominator of the educational improvements measure to those who begin an education component.

For the purpose of calculating overall state performance, OTA suggests that FNS not attempt to distinguish between satisfactory and outstanding educational components during the first 2 years after implementation of the performance standards. Asexperience with educational components accumulates, and as research now underway provides information on the benefits of education for the welfare population, FNS may want to reexamine the role of education within FSET.
The States may select one or more of three different categories of success.--OTA agrees with FNS that obtaining a high school diploma or GED certificate is an appropriate measure of educational improvement. OTA notes, however, that these two measures are different. GED certificate holders have passed a set of standardized tests, normed on graduating high school seniors. Those who qualify for a GED certificate have performed on the tests better than 25 percent or more of graduating high school seniors (see box A). Unlike the GED certificate, the criteria for awarding a high school diploma are not standardized and hence may vary across States and localities. On the other hand, high school diplomats have successfully completed required functions within an institutional setting that GED certificate holders have not.

The critical question is: how similar are holders of high school diplomas and of GED certificates with respect to the fundamental goal of FSET—attainingself-sufficiency, as measured by earnings? Published research is inadequate to answer this question. While a national survey of employers found that between 96 and 98 percent treated traditional high school graduates and holders of GED certificates the same in terms of starting salary, employment level, and opportunities for advancement, l11 

## whether these similarities continue with the passage of years is not clear. 112



## In view of the institutional history of the high school diploma and the scientific basis for the GED, OTA concurs with the use of either as a measure of educational improvement, unless further research demonstrates clear superiority of one over the other.

OTA takes issue with 64 hours of classroom training as the minimally acceptable level for the proposed measure of educational improvements. FNS’ justification is that 64 hours is a level of effort roughly comparable to that which a State agency would expend to help each participant find a job. 113 

## Sixty-four hours is also the number that results when 4 hours per week (the lower bound of Department of Education estimates that 4 to 6 hours per week are spent by adults in Adult Basic Education) is multiplied by 16 weeks (the median number of weeks that work registrants participate in the food stamp program).

Whether or not 64 hours is adequate time to make a significant gain in education is a function of many factors including the educational task that the learneris trying to master, the learner’s preexisting proficiency in the task, native ability, the number and spacing of the learnin g periods into which the 64 hours is divided, and the capability of the teacher, among other variables. For programs similar to those

11056 Federal Register 43173 (Aug. 30, 1991).

lll~~ew Malizio and Douglas Whitney, ‘‘Educational Credentials in Employment: A Nationwide Survey,’ American Council on Education and GED Research Brief, May 1985.

112~vestigator~ ~ttemp~g t. ~solve this issue: ~ &a&r, Rutg~s ufiversi~, ~~to~ NJ ad Stephen Cmeron ~d J~M HCX_ National Opinion Research Center at the University of Chicago, Chicago, IL.

11356 Federal Register 43173 (Aug. 30, 191).

### Section 4--Comparison Between FNS’ Proposed Standards and OTA’s Model Standards ● 

### 25



## Box A—What Is The GED?

GED stands for General Educational Development The GED tests are administered by the American Council on Education and are taken by people who have not completed high school but hope to demonstrate that they have knowledge equal to those who have high school diplomas. The GED tests have been used in all 

## 50 States since 1963. 1 

## In 1990 nearly 764,00 persons took the examination and 431,225 passed and were thereby certified as high school equivalents. 2 

## The percentage of those who are classified as high school graduates by virtue of the GED tests has increased substantially since the mid-50s. Of high school graduates in 1954, less than 2 percent were GED certified. 3 

## That number has risen to 16 percent today. 4



## The GED examination consists of 290 items in five test areas: literature and the arts, writing, mathematics, xamination is periodically revised to reflect the major content areas of high school science, and social studies. The e education and changes in how to best test whether applicants have achieved equivalent skills and knowledge. For example, in the last revision in 1988, the examination was changed to include a question requiring an essay answer and questions were changed to place more emphasis on problem-solving skills. Besides revision of test items to reflect changes in curricula and educational thought, equivalency depends on the more stringent standard of n orming the test through performance of those graduating from high school. The American Council on Education recommends that States using the GED tests set the base level for passing all five test areas at the level achieved by at least 25 percent of graduating high school seniors. Most States use a base of 30 percent or higher. 5



### %WOII V

. CttmerOn and Jamm J. HeckmaQ The 

### Nonequivalence of High School J!lqw”valents, report prepared tm tie Natio~ ~i~e

Fmmdation ad the Bureau of Labor Statistics, U.S. Department of Labor, IMember 1990, revised June 1991, p. 1.

@!D %* SeNice, 1990 SratMicaZReporr (WaahingtoXDC: Amerioan Council onlklucatiom 1990), 

## p. 2.

3c=mn @ *- op. cit., 

### footnote L P+ 4.

. 

## in which FSET participants would enroll, it is much trainees at the sixth grade reading level and easier to find general statements about program above improve their skill levels by at least one effectiveness than actual measurements of gains in and a half grade levels after only 20 hours of competence. 114 

## Information in the literature that study. Those below the sixth grade level bears on this point includes: progress more slowly, achieving a half year or

●

●

●

● 

## To improve a single grade level in a basic skill such as reading takes 100 hours on average. 115 

## In “exemplary” programs, participants can improve one grade level in 50 hours. 116 

## For those at a very rudimentary level of learning, such as being able to sound out words on road signs, it takes several hundred more hours of learning to be able to understand most newspaper stories. 117



## The USDA Graduate School’s Workforce 2000Skills Development Program, using individualized self-study in generic literacy areas, helped more improvement in reading grade level in 40 hours.

It is clear that reported learning in 64 hours varies widely. That is one reason that OTA finds that 64 hours of classroom tr aining is a poor measure of educational improvement. It is not the major reason, however. The greatest problem with designating completion of 64 hours of classroom training as a ‘‘success’ is that it is a participation measure rather than an outcome-based performance measure. As is clear from the 1988 Hunger Prevention Act, what is of interest is a measure of the 

## results of time spent

114Larry Mi.kulecky, Second Chance Basic Skills Education, Indiana University at BloomingtorL mimeo, no date, p. 228.

IM~s is me +~ hti mili~ recruits rind in large civiliau programs. Ibid., p. 226.
llGD~lfig, Ig84 in Mikulecky, op. cit., footnote 114, p. 226
llTHec~,

Op. cit., footnote 114, p. 226

### 26 ● 

### Performance 

## Standards for the Food Stamp Employment and Training Program



## Box B—What Kind 

## of Education Is Most Likely 

## To Help FSET Participants?

Research shows that what is taught in schools and in adult literacy programs differs from the reading, writing, and analytical tasks in the workplace. 1 

## The courts accept the distinction; tests of general reading ability for employment have been found unsuitable as job screening devices unless they reflect actual job reading demands (Griggs v. Duke Power). 2 

## Further, even highly literate men and women do not perform well with technical materkd on subjects with which they are unfamiliar, yet a low level of literacy maybe adequate with the same material if the reader is very familiar with the subject. Results of testing U.S. Navy personnel show that measured reading ability can be several grade levels below the difficulty of the material to be read, yet the material is understood if the reader is very familiar with the subject. 3



## The major distinction between the workplace and the classroom appears to be that in the workplace a person reads, writes, and computes-to-do, while in the classroom he or she reads, writes, and computes-to-learn, a distinction first made by Sticht from military research in 1975 and confirmed by Diehl and Mikulecky in 1980. 4 

## Put another way, students read texts to gather facts while workers read manuals, memos, schematics and other forms of material to do, to assess, and to solve problems 5 

## as well as to learn. 6 

## When reading to learn, students and workers use different strategies. Students take notes and re-read. Workers use a wider array of strategies including problem solving and relating new ideas to what they already know. 7 

## These differences probably contribute to the weak relationship between academic basic skills measured by standardized tests and job performance. 8



## A related issue is retention. Military recruits who learned basic skills with job-related materials retained most of what they learned, but more than half of the learning gains with material that was not job-related was lost within 8 weeks. 9 

## The literature does not indicate why learning with workplace material is more effective. Transfer, continued practice, and greater motivation are possible explanations for the greater effectiveness of learning with materials that are encountered during the workday. 10



## IW+S. a- of B&cation and TJ.s. I)epartmmt of Labor, ‘The Bottom Line: 

## Basic SkillS ill the WM#-” Was~_ X:

U*S. Gov ernrnent Prindng OHlce, 1988), p. 6. 

### %bid., p. 8.
31@xJw and fi~r, Trainin~The Competitive Edge, footnOte 7, p. 180.

4Paul V. De~@, “&Mic SHS ~~cation in Business and Jndustry: FactOH for Success Or Failure,” Contract paw submtt~ to
~ ~~ of ‘ibchnology Assessment Mtnuary 1990, p. 3.

%bid., p. 2.

6% HW~, ‘&Job ~tmcy:
~ MktiomMp Between SChOOl Preparation and Work@ace Acti%o” Re~4n8 Resea~c~

QuarferZy, VOL 17, no. 3,1982, p. 413.

71bid., p. 413.
8Dolkw, op. cit., 

## fmtnote, p. A. gTho~ G. Sti@~ in

~ wlecky, seco~ chance Basic Skills Education, Indiana thiversity at 

### BIOOd.W@NA *, no date, p. 240. I-lee@, Ibid., note 6, p. 

### 250.



## learning , or attempting to learn, rather than a Rather than 64 hours of classroom training or
measure of the time itself. meeting State-defined goals comparable to 64 hours of classroom training, OTA suggests that FNS For the same reasons that OTA disagrees with 64 substitute the following measure: One grade level or
hours of classroom training as a measure of educa- an equivalent competency-based educational functional level improvement in one of three basic tional improvement, OTA disagrees with allowing skills-reading, writing, or mathematics-as demStates to define goals comparable to participation in 64 hours of classroom training. onstrated by a widely used test. Examples of widely used normed achievement tests (which measure

### Section Comparison Between FNS’ Proposed Standards and OTA’s Model Standards ● 

### 27

1


## Early results from the Minority Female Single Parent (MFSP) demonstration study support the importance of relating basic skills acquisition to workplace tasks. This project, funded by the Rockefeller Foundation, provided education, training, job search assistance, and support services (e.g., child care) to minority mothers, many of whom were on welfare. The four MFSP projects differed significantly in how they provided remediation in basic skills and training in job skills. Two (in Providence and Atlanta) used a traditional sequential approach with remediation of basic skills required before participants could progress to skill training for specific jobs. One provided basic skills to the poorly prepared and gave general training g applicable to a variety of jobs in such subjects as electronics and mechanics to those with more skills. The fourth project, at the Center for Employment Training (CET) in San Jose, closely integrated basic skills and job skills training. Comparison of treatment and control groups clearly showed that only the integrated approach of the CET program significantly increased the percentage working and their average monthly wage. 11 

## The results of the MSFP projeet must be approached with caution-only four sites were studied and the success of CET may be due to factors such as its long history and its ties to its largely Hispanic community. The timeframe over which controls and participants have been compared (1 year after program entry) may also partly explain the superiority of the GET program. As mare time elapses, the sequential programs may prove to provide statistically significant earnings gains. Still, minority female single parents seem to value learning more if they can see that it is helping them to achieve their primary objective of getting a job. In the two sequential programs, many of the participants left during the required remedial basic educational classes and never began occupational training. These findings confirm an important principle of adult learning, or andragogy, as formulated by Malcolm Knowles. According to Knowles, the adult orientation to 1 earning is Me-or task-centered,rather than learning for the sake of learning. Hence, curricula should be organized around life situations (e.g., “Writing Better Business Utters”) rather than abstractions (e.g., courses such as “Composition I"). 12 

## For adults needing remediation of basic skills or English as a Second Language, relevance of what they are learning to the adult world of work is very important.

lljo~B-@ ~d -e Q@OQ J#~reJob~ and Hig&r Pay (New Yo~ ~: The Rockefeller Fmruk@m, 1990)J P“ ii”

12MAIrASrn

S. KMXVleS and Aasoeiates, Andragogy in Action (blew York NY: Jossey- BaSS Publkhers, 

### 1985), P. 12.



## grade level improvements) are the Test of Adult programs do not always correlate with the skills Basic Education (TABE) and the Adult BasicLearning Examination (ABLE), while the Comprehensive Adult Student Assessment System (CASAS)is a widely used competency based test instrument. 118 

## OTA suggests that each State be given the flexibility to select the test(s) that it will use to assess success in meeting the educational improvement measure, provided that the test(s) selected is widely used and nationally recognized. Research should continue to establish comparability among these

tests.

The measure is not tailored to the special needs of adult learners.— It is clear that strong basic skillsare important for a growing number of jobs. However, as shown in box B, the kind of academic basic skills typically offered in remedial educational needed to obtain and perform in good jobs.
Permitting the States to select competency basedtests as described above may encourage implementation of programs that integrate basic skills trainingwith occupational training and the consequent generation of additional data on the value of such combined training.

Initial National Standard
FNS’ proposed standard for State performance is success by 25 percent of FSET participants who complete educational components. The basis for this standard is expert opinion in light of insufficient data on which to base a standard. 119 

## OTA concurs that there is a lack of data and supports the proposed standard. OTA also supports FNS’ proposal to make adjustments for individual States in FY 1994 and

118~ow~ stich~ Te~ti~g andA~~e~~~ent in Adult Ba~iC &fuCati~n ~~ English as a second ~nguage program, R~II prepared for the U.S. Department of Edueatio@ Division of Adult Education and Literacy, January 1990.

11956 Federal Register 43174 (Aug. 30, 1991).

### 28 ● 

### Performance 

## Standards for the Food Stamp Employment and Training Program



## beyond if sufficient data become available to make such adjustments.



## Calculation of Overall Performance

To calculate overall State performance, OTA supports FNS’ proposal that each State be given a simple numerical rating-a 2 for outstanding, 1 foracceptable performance, O for unacceptable performance-for each of the outcome-based measures. 120 

## FNS defines acceptable performance as achieving the minimum performance standards (25 percent for job placements, $4.25 for average wages, 20 percent for food stamp case closures, and 25 percent success in educational improvements). FNS does not define ‘‘outstanding’ performance for these measures. OTA suggests that, in defining outstanding, FNS examine the 50th percentile of what the States achieved in recent years.
FNS proposes that the States not be required to be measured and ranked on educational improvements, but that they be able to win up to 2 extra points for outstanding programs. OTA would not distinguishbetween satisfactory and outstanding on this measure, allowing only 1 point. OTA would also exclude food stamp case closures from the measures and give extra weight (specifically, 1.5 times as much weight) for successful performance on the wage standard(e.g., a 3 for outstanding, 1.5 for acceptable performance, O for unacceptable performance). These ratings would be summed to give an overall rating for each State.
With four performance measures (three mandatory and one optional), and the above-mentioned rating scheme, FNS proposes to rate States earning less than 2 points as unsatisfactory, States with 2 to 4 points as satisfactory, and States with 5 or more points as outstanding. 121 

## Although OTA would only use two mandatory and one optional measure, OTA would, like FNS, require the States to meet the standard on at least two of the measures. Thus, OTA would rate States earning less than 2.5 points as unsatisfactory, States with 2.5 to 4 points as satisfactory, and States with 4.5 or more points as outstanding (See table 6).

FNS’ proposed ratings would not affect funding for the first 2 years of implementation of the new standards. Instead, incentive funds ($15 million of the total $75 million in State grants) would continueto be awarded on the basis of the current, participationbased performance standards. 122 

## Congress recently (in Public Law 102-237, signed December 13, 1991) directed FNS to change the participation standard from 50 percent of mandatory participants to 10 percent of such participants, and achievement of the new 10 percent standard will determine incentive awards for the next 2 fiscal years. Beginning in FY 1995, FNS proposes to award the $15 millionincentive funds on the basis of both State performance ratings and participation rates. 123



## Because the proposed perfo rmance measures (both OTA’s and FNS’) are only rough estimates of the effectiveness of State programs, OTA suggests that they not be used as the basis for financial rewards and sanctions. OTA recognizes that rewards and sanctions were mandated by Congress in the Food, Agriculture, Conservation and Trade Act of 1990 (Public Law 101-624). Nevertheless, Congress may wish to consider the possibility of using the new perfo rmance standards primarily to identify which States are in need of extra technical assistance andwhich States should be recognized with commendations or other nonmonetary rewards. Sanctionsmight then be reserved for persistent poor performance.

12056 Federa/&gister 43188 (Aug. 30, 1991).

12156 Federa/R@ter 43183 (Aug. 30, 191).

12256 Federa/RegiSter 43188 (Aug. 30, 1991).
lzs~id.

### Section 4--Comparison Between FNS’ Proposed Standards and OTA’s Model Standards .29

Table 6-Assessment of Overall Performance: Comparison of FNS’ Proposed Rule and OTAsS Model Standards

Measure FNS OTA

Job placement O points for unsatisfactory Same 1 point for satisfactory Same 2 points for outstanding Same

Average wage (FNS) O points for unsatisfactory Same Quarterly earnings (OTA 1 point for satisfactory 1.5 points 2 points for outstanding 3 points

Food stamp case closures O points for unsatisfactory Exclude this measure 1 point for satisfactory 2 points for outstanding

Educational improvements O points for unsatisfactory Same 1 point for satisfactory Same 2 points for outstanding (Does not distinguish between satisfactory and outstanding in the first 2 years) Formula for overall assessment

FNS OTA
Job placement score+ wage score + food stamp Job placement score + wage score + educacase closure score + educational improvement tional improvement score (if the State offers score (if the State offers education) education)

Unsatisfactory performance Less than 2 points Less than 2.5 points

Satisfactory performance 2 to 4 points 2.5 to 4.5 points

Outstanding performance Over 4 points Over 4.5 points

# Section 5 

# Policy Options

As currently structured, FSET is not costeffective. The taxpayers’ costs to administer the program are offset neither by decreases in the food stamp rolls nor by increased income taxes from increased employment of food stamp recipients. The following two options, taken together or separately, might improve the effectiveness of FSET.



# OPTION 1: MAKE FSET PARTICIPATION VOLUNTARY

There are at least two reasons to consider making FSET voluntary: 1) the mandatory aspects of FSET are only weakly enforced, and have little impact when they are; 2) volunteers might be more likely to complete and benefit from FSET.

As discussed in section 3, FNS has allowed the States to grant categorical and personal exemptions to about 30 percent of work registrants in recent years. The States were expected to either enroll or initiate sanctions against 50 percent of the remaining “nonexempt mandatory” participants. The large fraction of exemptions raises important questions about equity. Do caseworkers throughout the Nation follow uniform guidelines when deciding which medical and transportation problems are serious enough to warrant exemptions? Under the new proposal allowing the States to enroll as few as 10 percent of work registrants, these equity questions may become even more serious. The few work registrants who are required to participate may resent the others who are not required to do so.
There are two possible approaches to making the program more equitable while targeting it to a smaller group. If Congress wishes to keep the program mandatory, the States could use a simple queuing system, so that 10 percent of non-exempt mandatory participants were enrolled at any given time, while the others were on a waiting list. Duringtheir time on the waiting list, non-exempt mandatories would not face the threat of sanctions. Many— presumably those who are most employable--might find jobs and leave the food stamp rolls” without the need of FSET. Those work registrants who reached the top of the waiting list and were called in for an initial interview with an FSET counselor would face the threat of sanctions if they did not appear or if they subsequently dropped out of the program. 124



## A second approach would be to make FSET voluntary. Among those who are not exempt, the threat of sanctions appears to have little impact. During FY 1988, 34 percent of this group never appeared for their first interview with an FSET caseworker. 125 

## Many of these no shows as well as those who enrolled but later dropped out were able to cure their noncompliance through repeated phone calls, and continued to receive food stamps. Many local food stamp agencies—about one-third of those surveyed in FY 1988—reported they would “try anything to avoid sanctioning a client. ’’ 126 

## Even when a Notice of Adverse Action is sent out immediately, processing and required notification time result in very few food stamp recipients having their food stamps terminated. This is because the average work registrant is only on the food stamp rolls for 3 months, and it often takes 1 or 2 months before a registrant is denied recertification for food stamps. 127 Thus, 

## sanctioned registrants frequently forfeit no, or only 1 month, of benefits.

Those who fail to comply with FSET requirements may do so for the very good reason that they can find a job without the program. Among workregistrants deemed to be mandatory FSET participants in FY 1988, 56 percent of the no shows wereemployed 1 year after random assignment, compared to 51 percent of those who participated in FSET. 128



## The mandatory aspect of FSET derives from its origins in workfare. As an optional activity within FSET, workfare has never been very attractive to

l~D~el Frie~d~, MDRC, Inc., personal communieatiow N OV . 25, 1991.
l~w, et al., op. cit., footnote 15, pp. 6-8.

1261bid., pp. 3-8.

127Charles Usheret al., “Long-’IkrmParticipationin the Food Stamp Program by WorkRegistrants, ‘‘ in Carole Trippeet aL, FoodStamp Policy Issues: ResuZtsfiom Recent Research (Alexan&i~ VA: U.S. Department of Agriculture, Food and Nutrition Service, 1990), p. 199.
IXJ-, et aI., op. cit., footnote 15, p. 6-26.

--l

### 32 ● 

### Performance 

## Standardsfor the Food Stamp Employment and Training Program



## local FSAs 129 

## primarily because the relatively small value of the food stamp benefits received by mosthouseholds requires creation of workfare jobs lasting only 7 or 8 hours a week. 130 

## However, the idea of equating required participation with the value of the food stamp benefits continues in the FSET program. As a result, participants are only required to spend only a minimal time--e.g., about 24 hours in the District of Columbia-in the program.
Despite its ineffectiveness, the mandatory aspect of FSET requires a great deal of administrative effort--40 percent of staff time in FY 1988. By comparison, employability assessments and direct provision of employment and training services accounted for only 18 percent of staff time. 131 

## Assigning the limited FSET funds to a smaller group of volunteers might be far more effective than the current approach of spreading the money thinly

across a 

## large group of participants and nonparticipants (who require administrative funds to be tracked and sanctioned).
Relying on volunteers need not result in Stateagencies creaming the most employable work registrants. By giving extra credit 

## for job placements, wages, and educational improvements for the hard to employ, FNS could encourage the States to make extra efforts to attract HTE participants to volunteer.
Would anyone appear if FSET were made voluntary? OTA notes that during 1988, 150,000 foodstamp recipients who were not required to participate did so. This is over half the number (240,000) who might be enrolled if the States choose to enroll only the new required minimum of 10 percent. It is approximately equal to a 6 percent participation rate. It would seem unreasonable to set a participation rate for a 

## voluntary program. However, if this option were chosen, Congress and FNS might want to monitor participation to insure that it not fall to minimal levels. Over time, participation rates ought to grow if the 

## States develop effective programs that are attractive to food stamp recipients. 

# OPTION 2: MERGE FSET INTO JTPA

To reduce administrative costs and increase direct provision of employment and training services, Congress could consider merging FSET into JTPA. 132

## During FY 1988, three-quarters of all FSA’s coordinated some or all of their FSET activities with JTPA. Some FSAs simply referred clients to JTPA, while others contracted for specific services such asemployability assessments. Both types of arrangements require using limited Federal funds to support administrative costs of two separate agencies. And, for food stamp recipients, the simple fact of having to go to more than one office maybe discouraging. As one observer put it, ‘‘the more disadvantaged the client, the more likely she or he will fall through the fissures of program fragmentation.’’1 33



## The National Commission for Employment Policy (NCEP), a 

## presidentially appointed body, recently made a similar ‘merge” recommendation to the President and the Congress. Following an extensive process of research and consultation at the

## Federal, State, and local levels, NCEP recommended that 

## JTPA, JOBS, FSET, and all other federally funded employment and training programs aimed at disadvantaged persons be merged into a 

## single Federal program operated by a 

## single Federalagency. The rationale for the proposed reorganization included minimizing conflicting regulations,improving program management, reducing administrative costs, and enabling the “States to deal with fewer contact points in Washington. ’’134

FSET was created in part because of concern that food stamp recipients were not receiving the employment and training services they needed from other agencies. Specifically, the State employment security agencies which administered the work registration program provided little assistance to work registrants and placed only a very small

1zq3Y lg85, 4 ~ems ~M wor~m ~~e av~ble, Ordy 19 out of 3,ooO local jurisdictions operated such programs. Center on Budget ~d po@ l%orities, Food Stamp Employment and Training Resource Guide (1%.shingtoq DC: 1986), p. 24.

IW.S. Congress, House Committee on Agriculture, House Report No. 955-464, p. 2301.
131*, et al., op. cit., footnote 15, p. ix. lsz~otherpossibilityw ould &tomerge~ETwithJOBs. This option is 

## apxbecausebotiprograrna areusuWy administered by the me Stite and local welfare agencies. However, as discussed in section 2, the clients served by JOBS differ from those served by FSET. The FSET clientele is more similar to that served by JTFA.
133Robert E. Friem “S~te H~ Investment Policy: Just on the Horizon,” EntrepreneurialEconomy Review, autumn 1990, p. 23.

l~Johu C. Gartland,

~ NCEP, letter to the PresidenL Sept. 30, 1991.

### Section 5--Policy Options .33



## fraction (less were merged occur? than 10 percent) into jobs. 135 

## If FSET

into JTPA, would the same problem

JTPA’s previous performance standards, emphasizing m aximum job placements at minimum cost, caused some local program operators to reduce services to welfare recipients, focusing on more employable people. 136 

## However, JTPA’s new performance standards for program years 1990-92 137 

## are designed to increase service to the hard-to-serve and enhance long-term employability.
If Congress chose to merge the two programs, 138 

## it could encourage service to food stamp recipients through a contracting arrangement between USDA and the U.S. Department of Labor (DOL). The addition of FSET funds would add about 8 percent to JTPA’s Title II-A budget, which DOL could use to serve an additional 8 percent of clients, or 96,000 food stamp recipients, ass uming the JTPA Title IIa average cost of $1,646 per participant. There is a precedent for delivering employment and training services needed by other Federal agencies through DOL. In the National Defense Authorization Act for FY 1991 (Public Law 101-510, approved October 1990), Congress approved the transfer of $150 million from the Department of Defense to DOL to provide retraining and reemployment services to laid-off defense workers. 139



## The number of food stamp clients served under such an arrangement (96,000) would be much smaller than the 1990 figure of 1.35 million served by FSET plus about 120,000 served by JTPA. However, even if the programs remained separate, the new 10 percent participation standard can beexpected to greatly reduce the number of participants in FSET to as few as 240,000 people.

135Camil Associates, “Semices to Applicants Required To Be Registered With the U.S. Employment Semice,” Report prepared for the Office of Policy, Evaluation and Research U.S. Department of Labor, Employment and ‘Mining Adrmms “ “ tratioq Washingbn, DC. 1979.

136Di&smoq et & op. cit., foo~ote 38, p. 194. N~o~y, JTP” sewed we~~ ~ipien~ in propofion to theh 

## representation in the. elQible population in 1985; however, school dropouts wereunderserved and received little remedial educatioq and clients who were “less job ready” (including welfare recipients) received less occupational training than those who were more job ready. U.S. Congress, General Accounting OffIce, Job Training Partnership Act: Sem”ces and Outcomes for Pati”cipants With Di#ering Needs (Washington DC: 1989), p. 3.

137u.s. Department of Labor, Employment and Trq

~ “ “stratiow “Notice of performance standards for PY 1990-1991,” Federal Register, vol. 55, No. 72 (Apr. 13, 1990), p. 14016. 138A comprehensive ~tio~ ev~mtion of ~A, involving 30,000 individ~s in ~n@ol ~d expefient groups, is now underway. The K%Xlh, expected in the summer of 1992, will be useful if Congress considers the possibility of merging FSET and JTPA.

13~.s. conwe55, OT&&fer the cold war: Living with Lo~erDefe~e spending, o~-rrE-524 @C: us. Gm, J~~ 1